Enciclopedia de Dispositivos Médicos e Instrumentación

ENCYCLOPEDIA OF
MEDICAL DEVICES AND

INSTRUMENTATION
Second Edition
VOLUME 4
Hydrocephalus, Tools for Diagnosis and Treatment of Monoclonal Antibodies
ENCYCLOPEDIA OF MEDICAL DEVICES AND INSTRUMENTATION, SECOND EDITION

Editor-in-Chief
John G. Webster
University of WisconsinMadison
Editorial Board
David Beebe
Jerry M. Calkins
University of Arizona College of Medicine
Michael R. Neuman
Michigan Technological University
Joon B. Park
University of Iowa
Edward S. Sternick
TuftsNew England Medical Center
Editorial Staff
Vice President, STM Books: Janet Bailey
Associate Publisher: George J. Telecki
Editorial Director: Sean Pidgeon
Director, Book Production and Manufacturing:
Camille P. Carter
Production Manager: Shirley Thomas
Illustration Manager: Dean Gonzalez
Senior Production Editor: Kellsee Chu
Editorial Program Coordinator: Surlan Murrell
ENCYCLOPEDIA OF
MEDICAL DEVICES AND

INSTRUMENTATION
Second Edition
Volume 4
Hydrocephalus, Tools for Diagnosis and Treatment of Monoclonal Antibodies
Edited by
John G. Webster
The Encyclopedia of Medical Devices and Instrumentation is available online at

http://www.mrw.interscience.wiley.com/emdi
A John Wiley & Sons, Inc., Publication
Copyright # 2006 by John Wiley & Sons, Inc. All rights reserved.
Published by John Wiley & Sons, Inc., Hoboken, New Jersey
Published simultaneously in Canada
No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic,
mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 of the 1976 United States
Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate
per-copy fee to the Copyright Clearance Center, Inc., 222, Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978)
750-4470, or on the web at www.copyright.com. Requests to the Publisher for permission should be addressed to the Permissions
Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at
http://www.wiley.com/go/permission.
Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they
make no representations or warrnaties with respect to the accuracy or completeness of the contents of this book and specifically disclaim
any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales
representatives or written sales materials. The advice and strategies contained herein may not be suitable for your situation. You should
consult with a professional where appropriate. Neither the publisher nor author shall be liable for any loss of profit or any other commercial
damages, including but not limited to special, incidental, consequential, or other damages.
For general information on our other products and services or for technical support, please contact our Customer Care Department
within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317) 572-4002.
Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in
electronic formats. For more information about Wiley products, visit our web site at www.wiley.com.
Library of Congress Cataloging-in-Publication Data:

Library of Congress Cataloging-in-Publication Data
Encylopedia of medical devices & instrumentation/by John G. Webster,
editor in chief. 2nd ed.
p. ; cm.
Rev. ed. of: Encyclopedia of medical devices and instrumentation. 1988.
Includes bibliographical references and index.
ISBN-13 978-0-471-26358-6 (set : cloth)
ISBN-10 0-471-26358-3 (set : cloth)
ISBN-13 978-0-470-04069-0 (v. 4 : cloth)
ISBN-10 0-470-04069-6 (v. 4 : cloth)
1. Medical instruments and apparatusEncyclopedias. 2. Biomedical engineeringEncyclopedias. 3. Medical physicsEncyclopedias.
4. MedicineData processingEncyclopedias. I. Webster, John G.,
1932- . II. Title: Encyclopedia of medical devices and instrumentation.
[DNLM: 1. Equipment and SuppliesEncyclopediasEnglish. W 13
E555 2006]
R856.A3E53 2006
610.2803dc22
2005028946
Printed in the United States of America

10 9 8 7 6 5 4 3 2 1
CONTRIBUTOR LIST
ABDEL HADY, MAZEN, McMaster University, Hamilton,
Ontario Canada, Bladder Dysfunction, Neurostimulation
of
ABEL, L.A., University of Melbourne, Melbourne, Australia,
Ocular Motility Recording and Nystagmus
ABREU, BEATRIZ C., Transitional Learning Center at Galveston, Galveston, Texas, Rehabilitation, Computers in
Cognitive
ALEXANDER, A.L., University of WisconsinMadison, Madison, Wisconsin, Magnetic Resonance Imaging
ALI, ABBAS, University of Illinois, at Urbana-Champaign,
Bioinformatics
ALI, MUFTU, School of Dental Medicine, Boston, Massachusetts, Tooth and Jaw, Biomechanics of
ALPERIN, NOAM, University of Illinois at Chicago, Chicago,
Illinois, Hydrocephalus, Tools for Diagnosis and Treatment of
ANSON, DENIS, College Misericordia, Dallas, Pennsylvania,
Environmental Control
ARENA, JOHN C., VA Medical Center and Medical College of
Georgia, Biofeedback
ARIEL, GIDEON, Ariel Dynamics, Canyon, California, Biomechanics of Exercise Fitness
ARMSTRONG, STEVE, University of Iowa, Iowa City, Iowa,
Biomaterials for Dentistry
ASPDEN, R.M., University of Aberdeen, Aberdeen, United
Kingdom, Ligament and Tendon, Properties of
AUBIN, C.E., Polytechniquie Montreal, Montreal Quebec,
Canada, Scoliosis, Biomechanics of
AYRES, VIRGINIA M., Michigan State University, East Lansing, Michigan, Microscopy, Scanning Tunneling
AZANGWE, G., Ligament and Tendon, Properties of
BACK, LLOYD H., California Institute of Technology, Pasadena, California, Coronary Angioplasty and Guidewire
Diagnostics
BADYLAK, STEPHEN F., McGowan Institute for Regenerative
Medicine, Pittsburgh, Pennsylvania, Sterilization of Biologic Scaffold Materials
BANDYOPADHYAY, AMIT, Washington State University, Pullman, Washington, Orthopedic Devices, Materials and
Design for
BANERJEE, RUPAK K., University of Cincinnati, Cincinnati,
Ohio, Coronary Angioplasty and Guidewire Diagnostics
BARBOUR, RANDALL L., State University of New York Downstate Medical Center, Brooklyn, New York, Peripheral
Vascular Noninvasive Measurements
BARKER, STEVEN J., University of Arizona, Tucson, Arizona,
Oxygen Monitoring
BARTH, ROLF F., The Ohio State University, Columbus,
Ohio, Boron Neutron Capture Therapy
BECCHETTI, F.D., University of Michigan, Ann Arbor, Michigan, Radiotherapy, Heavy Ion
BELFORTE, GUIDO, Politecnico di Torino Department of
Mechanics, Laryngeal Prosthetic Devices
BENKESER, PAUL, Georgia Institute of Technology, Atlanta,
Georgia, Biomedical Engineering Education
BENNETT, JAMES R., University of Iowa, Iowa City, Iowa,
Digital Angiography
BERSANO-BEGEY, TOMMASO, University of Michigan, Ann

Arbor, Michigan, Microbioreactors
BIGGS, PETER J., Harvard Medical School, Boston, Massachusetts, Radiotherapy, Intraoperative
BIYANI, ASHOK, University of Toledo, and Medical College of
Ohio, Toledo, Ohio, Human Spine, Biomechanics of
BLOCK, W.F., University of WisconsinMadison, Madison,
Wisconsin, Magnetic Resonance Imaging
BLUE, THOMAS E., The Ohio State University, Columbus,
Ohio, Boron Neutron Capture Therapy
BLUMSACK, JUDITH T., Disorders Auburn University,
Auburn, Alabama, Audiometry
BOGAN, RICHARD K., University of South Carolina, Columbia, South Carolina, Sleep Laboratory
BOKROS, JACK C., Medical Carbon Research Institute, Austin, Texas, Biomaterials, Carbon
BONGIOANNINI, GUIDO, ENT Division Mauriziano Hospital,
Torino, Italy, Laryngeal Prosthetic Devices
BORAH, JOSHUA, Applied Science Laboratories, Bedford,
Massachusetts, Eye Movement, Measurement Techniques for
BORDEN, MARK, Director of Biomaterials Research, Irvine,
California, Biomaterials, Absorbable
BORTON, BETTIE B., Auburn University Montgomery, Montgomery, Alabama, Audiometry
BORTON, THOMAS E., Auburn University Montgomery, Montgomery, Alabama, Audiometry
BOSE SUSMITA,, Washington State University, Pullman,
Washington, Orthopedic Devices, Materials and Design
for
BOVA, FRANK J., M. D. Anderson Cancer Center Orlando,
Orlando, FL, Radiosurgery, Stereotactic
BRENNER, DAVID J., Columbia University Medical Center,
New York, New York, Computed Tomography Screening
BREWER, JOHN M., University of Georgia, Electrophoresis
BRIAN, L. DAVIS, Lerner Research Institute, The Cleveland
Clinic Foundation, Cleveland, Ohio, Skin, Biomechanics
of
BRITT, L.D., Eastern Virginia Medical School, Norfolk, Virginia, Gastrointestinal Hemorrhage
BRITT, R.C., Eastern Virginia Medical School, Norfolk,
Virginia, Gastrointestinal Hemorrhage
BROZIK, SUSAN M., Sandia National Laboratories, Albuquerque, New Mexico, Microbial Detection Systems
BRUNER, JOSEPH P., Vanderbilt University Medical
Center, Nashville, Tennessee, Intrauterine Surgical
Techniques
BRUNSWIG NEWRING, KIRK A., University of Nevada, Reno,
Nevada, Sexual Instrumentatio n
BRUYANT, PHILIPPE P., University of Massachusetts, North
Worcester, Massachusetts, Nuclear Medicine, Computers
in
BUNNELL, BERT J., Bunnell Inc., Salt Lake City, Utah, High
Frequency Ventilation
CALKINS, JERRY M., Defense Research Technologies, Inc.,
Rockville, Maryland, Medical Gas Analyzers
CANNON, MARK, Northwestern University, Chicago, Illinois,
Resin-Based Composites
v
vi
CONTRIBUTOR LIST
CAPPELLERI, JOSEPH C., Pfizer Inc., Groton, Connecticut,

Quality-of-Life Measures, Clinical Significance of
CARDOSO, JORGE, University of Madeira, Funchal, Portugal,
Office Automation Systems
CARELLO, MASSIMILIANA, Politecnicodi Torino Department
of Mechanics, Laryngeal Prosthetic Devices
CASKEY, THOMAS C., Cogene Biotech Ventures, Houston,
Texas, Polymerase Chain Reaction
CECCIO, STEVEN, University of Michigan, Ann Arbor, Michigan, Heart Valve Prostheses, In Vitro Flow Dynamics of
CHAN, JACKIE K., Columbia University, New York, New
York, Photography, Medical
CHANDRAN, K.B., University of Iowa, Iowa City, Iowa, Heart
Valve Prostheses
CHATZANDROULIS, S., NTUA, Athens, Attiki, Greece, Capacitive Microsensors for Biomedical Applications
CHAVEZ, ELIANA, University of Pittsburgh, Pittsburgh, Pennsylvania, Mobility Aids
CHEN, HENRY, Stanford University, Palo Alto, California,
Exercise Stress Testing
CHEN, JIANDE, University of Texas Medical Branch, Galveston, Texas, Electrogastrogram
CHEN, YAN, Lerner Research Institute, The Cleveland Clinic
Foundation, Cleveland, Ohio, Skin, Biomechanics of
CHEYNE, DOUGLAS, Hospital for Sick Children Research
Institute, Biomagnetism
CHUI, CHEN-SHOU, Memorial Sloan-Kettering Cancer Center, New York, New York, Radiation Therapy Treatment
Planning, Monte Carlo Calculations in
CLAXTON, NATHAN S., The Florida State University, Tallahassee, Florida, Microscopy, Confocal
CODERRE, JEFFREY A., Massachus etts Institute of Technology, Cambridge, Massachusetts, Boron Neutron Capture
Therapy
COLLINS, BETH, University of Mississippi Medical Center,
Jackson, Mississippi, Hyperbaric Medicine
COLLINS, DIANE, University of Pittsburgh, Pittsburgh, Pennsylvania, Mobility Aids
CONSTANTINOU, C., Columbia University Radiation Oncology, New York, New York, Phantom Materials in Radiology
COOK, ALBERT, University of Alberta, Edmonton, Alberta,
Canada, Communication Devices
COOPER, RORY, University of Pittsburgh, Pittsburgh, Pennsylvania, Mobility Aids
CORK, RANDALL C., Louisiana State University,
Shreveport, Louisiana, Monitoring, Umbilical Artery
and Vein, Blood Gas Measurements; Transcuta neous
Electrical Nerve Stimulation (TENS); Ambulatory
Monitoring
COX, JOSEPHINE H., Walter Reed Army Institute of Research,
Rockville, Maryland, Blood Collection and Processing
CRAIG, LEONARD, Feinberg School of Medicine of Northwestern University, Chicago, Illinois, Ventilators, Acute Medical Care
CRESS, CYNTHIA J., University of Nebraska, Lincoln,
Nebraska, Communicative Disorders, Computer Applications for
CUMMING, DAVID R.S., University of Glasgow, Glasgow,
United Kingdom, Ion-Sensitive Field-Effect Transistors
CUNNINGHAM, JOHN R., Camrose, Alberta, Canada, Cobalt 60
Units for Radiotherapy
DALESSANDRO, DAVID, Montefiore Medical Center, Bronx,
New York, HeartLung Machines
DAMBRA, MICHAEL N., Harvard Medical School, Cambridge,

Massachusetts, Cardiac Output, Thermodilution Measurement of
DADSETAN, MAHROKH, Mayo Clinic, College of Medicine,
Rochester, Minnesota, Microscopy, Electron
DALEY, MICHAEL L., The University of Memphis, Memphis,
Tennessee, Monitoring, Intracranial Pressure
DAN, LOYD, Linkoping University, Linkoping, Sweden, Thermocouples
DAS, RUPAK, University of Wisconsin, Madison, Wisconsin,
Brachytherapy, High Dosage Rate
DATTAWADKAR, AMRUTA M., University of Wisconsin,
Madison, Madison, Wisconsin, Ocular Fundus Reflectometry
DAVIDSON, MICHAEL W., The Florida State University, Tallahassee, Florida, Microscopy, Confocal
DE LUCA, CARLO, Boston University, Boston, Massachusetts,
Electromyography
DE SALLES, ANTONIO A.F., UCLA Medical School, Los
Angeles, California, Stereotactic Surgery
DECAU, SABIN, University of Maryland, School of Medicine,
Shock, Treatment of
DECHOW, PAUL C., A & M University Health Science Center,
Dallas, Texas, Strain Gages
DELBEKE, JEAN, Catholique University of Louvain, Brussels,
Belgium, Visual Prostheses
DELLOSSO, LOUIS F., Case Western Reserve University,
Cleveland, Ohio, Ocular Motility Recording and Nystagmus
DELORME, ARNAUD, University of San Diego, La Jolla, California, Statistical Methods
DEMENKOFF, JOHN, Mayo Clinic, Scottsdale, Arizona, Pulmonary Physiology
DEMIR, SEMAHAT S., The University of Memphis and The
University of Tennessee Health Science Center, Memphis,
Tennessee, Electrophysiology
DEMLING, ROBERT H., Harvard Medical School, Skin Substitute for Burns, Bioactive
DENNIS, MICHAEL J., Medical University of Ohio, Toledo,
Ohio, Computed Tomography
DESANTI, LESLIE, Harvard Medical School, Skin Substitute
for Burns, Bioactive
DEUTSCH, STEVEN, Pennsylvania State University, University Park, Pennsylvania, Flowmeters
DEVINENI, TRISHUL, Conemaugh Health System, Biofeedback
DI BELLA EDWARD, V.R., University of Utah, Tracer Kinetics
DIAKIDES, NICHOLAS A., Advanced Concepts Analysis, Inc.,
Falls Church, Virginia, Thermography
DOLAN, PATRICIA L., Sandia National Laboratories, Albuquerque, New Mexico, Microbial Detection Systems
DONOVAN, F.M., University of South Alabama, Cardiac Output, Indicator Dilution Measurement of
DOUGLAS, WILSON R., Childrens Hospital of Philadelphia,
Philadelphia, Pennsylvania, Intrauterine Surgical Techniques
DRAPER, CRISSA, University of Nevada, Reno, Nevada, Sexual Instrumentation
DRZEWIECKI, TADEUSZ M., Defense Research Technologies,
Inc., Rockville, Maryland, Medical Gas Analyzers
DURFEE, W.K., University of Minnesota, Minneapolis, Minnesota, Rehabilitation and Muscle Testing
DYRO, JOSEPH F., Setauket, New York, Safety Program,
Hospital
CONTRIBUTOR LIST
DYSON, MARY, Herts, United Kingdom, Heat and Cold,

Therapeutic
ECKERLE, JOSEPH S., SRI International, Menlo Park, California, Tonometry, Arterial
EDWARDS, BENJAMIN, University of Wisconsin-Madison,
Madison, Wisconsin, Codes and Regulations: Radiation
EDWARDS, THAYNE L., University of Washington, Seattle,
Washington, Chromatography
EKLUND, ANDERS, University of Illinois at Chicago, Chicago,
EL SOLH, ALI A., Erie County Medical Center, Buffalo, New
York, Sleep Studies, Computer Analysis of
ELMAYERGI, NADER, McMaster University, Hamilton,
Ontario, Canada, Bladder Dysfunction, Neurostimulation of
ELSHARYDAH, AHMAD, Louisiana State University, Baton
Rouge, Louisiana, Ambulatory Monitoring; Monitoring,
Umbilical Artery and Vein, Blood Gas Measurements
FADDY, STEVEN C., St. Vincents Hospital, Sydney, Darlinghurst, Australia, Cardiac Output, Fick Technique for
FAHEY, FREDERIC H., Childrens Hospital Boston, Computed
Tomography, Single Photon Emission
FAIN, S.B., University of WisconsinMadison, Madison,
Wisconsin, Magnetic Resonance Imaging
FELDMAN, JEFFREY, Childrens Hospital of Philadelphia,
Philadelphia, Pennsylvania, Anesthesia Machines
FELLERS, THOMAS J., The Florida State University, Tallahassee, Florida, Microscopy, Confocal
FERRARA, LISA, Cleveland Clinic Foundation, Cleveland,
Ohio, Human Spine, Biomechanics of
FERRARI, MAURO, The Ohio State University, Columbus,
Ohio, Drug Delivery Systems
FONTAINE, ARNOLD A., Pennsylvania State University, University Park, Pennsylvania, Flowmeters
FOUST, MILTON J., JR, Medical University of South Carolina
Psychiatry and Behavioral Sciences, Charleston, South
Carolina, Electroconvulsive Therapy
FRASCO, PETER, Mayo Clinic Scottsdale, Scottsdale, Arizona,
Temperature Monitoring
FRAZIER, JAMES, Louisiana State University, Baton Rouge,
Louisiana, Ambulatory Monitoring
FREIESLEBEN DE BLASIO, BIRGITTE, University of Oslo, Oslo,
Norway, Impedance Spectroscopy
FRESTA, MASSIMO, University of Catanzaro Magna Grcia,
Germaneto (CZ), Italy, Drug Delivery Systems
FREYTES, DONALD O., McGowan Institute for Regenerative
Medicine, Pittsburgh Pennsylvania, Sterilization of Biologic Scaffold Materials
FROEHLICHER, VICTOR, VA Medical Center, Palo Alto, California, Exercise Stress Testing
FUNG, EDWARD K., Columbia University, New York,
New York, Photography, Medical
GAGE, ANDREW A., State University of New York at Buffalo,
Buffalo, New York, Cryosurgery
GAGLIO, PAUL J., Columbia University College of Physicians
and Surgeons, Liver Transplantation
GARDNER, REED M., LDS Hospital and Utah University, Salt
Lake City, Utah, Monitoring, Hemodynamic
GEJERMAN, GLEN, Hackensack University Medical, Hackensack, New Jersey, Radiation Therapy, Quality Assurance
in
GEORGE, MARK S., Medical University of South Carolina
Psychiatry and Behavioral Sciences, Charleston, South
Carolina, Electroconvulsive Therapy
vii
GHARIEB, R.R., Infinite Biomedical Technologies, Baltimore,

Maryland, Neurological Monitors
GLASGOW, GLENN P., Loyola University of Chicago, Maywood, Illinois, Radiation Protection Instrumentation
GLASGOW, GLENN, University of Wisconsin-Madison, Madison, Wisconsin, Codes and Regulations: Radiation
GOEL, VIJAY K., University of Toledo, and Medical College of
Ohio, Toledo, Ohio, Human Spine, Biomechanics of
GOETSCH, STEVEN J., San Diego Gamma Knife Center, La
Jolla, California, Gamma Knife
GOLDBERG, JAY R., Marquette University Milwaukee, Wisconsin, Minimally Invasive Surgery
GOLDBERG, ZELENNA, Department of Radiation Oncology,
Davis, California, Ionizing Radiation, Biological Effects
of
GOPALAKRISHNAKONE, P., National University of Singapore,
Singapore, Immunologically Sensitive Field-Effect Transistors
GOPAS, JACOB, Ben Gurion University of the Negev, Beer
Sheva, Israel, Monoclonal Antibodies
GORGULHO, ALESSANDRA, UCLA Medical School, Los
Angeles, California, Stereotactic Surgery
GOUGH, DAVID A., University of California, La Jolla, California, Glucose Sensors
GOUSTOURIDIS, D., NTUA, Athens, Attiki, Greece, Capacitive
Microsensors for Biomedical Applications
GRABER, HARRY L., State University of New York Downstate
Medical Center, Brooklyn, New York, Peripheral Vascular
Noninvasive Measurements
GRACA, M., Louisiana State University, Baton Rouge,
Louisiana, Boron Neutron Capture Therapy
GRANT, WALTER III, Baylor College of Medicine, Houston,
Texas, Radiation Therapy, Intensity Modulated
GRAYDEN, EDWARD, Mayo Health Center, Albertlea, Minnesota, Cardiopulmonary Resuscitation
GREEN, JORDAN R., University of Nebraska, Lincoln,
Nebraska, Communicative Disorders, Computer Applications for
HAEMMERICH, DIETER, Medical University of South Carolina,
Charleston, South Carolina, Tissue Ablation
HAMAM, HABIB, Universite de Moncton, Moncton New Brunswick, Canada, Lenses, Intraocular
HAMMOND, PAUL A., University of Glasgow, Glasgow, United
Kingdom, Ion-Sensitive Field-Effect Transistors
HANLEY, JOSEPH, Hackensack University Medical, Hackensack, New Jersey, Radiation Therapy, Quality Assurance
in
HARLEY, BRENDAN A., Massachusetts Institute of Technology, Skin Tissue Engineering for Regeneration
HARPER, JASON C., Sandia National Laboratories, Albuquerque, New Mexico, Microbial Detection Systems
HASMAN, ARIE, Maastricht, The Netherlands, Medical Education, Computers in
HASSOUNA, MAGDY, Toronto Western Hospital, Toronto,
Canada, Bladder Dysfunction, Neurostimulation of
HAYASHI, KOZABURO, Okayama University of Science,
Okayama, Japan, Arteries, Elastic Properties of
HENCH, LARRY L., Imperial College London, London, United
Kingdom, Biomaterials: Bioceramics
HETTRICK, DOUGLAS A., Sr. Principal Scientist Medtronic,
Inc., Minneapolis, Minnesota, Bioimpedance in Cardiovascular Medicine
HIRSCH-KUCHMA, MELISSA, University of Central Florida
NanoScience Technology Center, Orlando, Florida,
Biosurface Engineering
viii
CONTRIBUTOR LIST
HOLDER, GRAHAM E., Moorfields Eye Hospital, London,

United Kingdom, Electroretinography
HOLMES, TIMOTHY, St. Agnes Cancer Center, Baltimore,
Maryland, Tomotherapy
HONEYMAN-BUCK, JANICE C., University of Florida, Gainesville, Florida, Radiology Information Systems
HOOPER, BRETT A., Arete Associates, Arlington, Virginia,
Endoscopes
HORN, BRUCE, Kaiser Permanente, Los Angeles, California,
X-Rays Production of
HORNER, PATRICIA I., Biomedical Engineering Society
Landover, Maryland, Medical Engineering Societies
and Organizations
HOROWITZ, PAUL M., University of Texas, San Antonio,
Texas, Fluorescence Measurements
HOU, XIAOLIN, Ris National Laboratory, Roskilde, Denmark, Neutron Activation Analysis
HOVORKA, ROMAN, University of Cambridge, Cambridge,
United Kingdom, Pancreas, Artificial
HUANG, H.K., University of Southern California, Teleradiology
HUNT, ALAN J., University of Michigan, Ann Arbor, Michigan, Optical Tweezers
HUTTEN, HELMUT, University of Technology, Graz, Australia,
Impedance Plethysmography
IAIZZO, P.A., University of Minnesota, Minneapolis, Minnesota, Rehabilitation and Muscle Testing
IBBOTT, GEOFFREY S., Anderson Cancer Center, Houston,
Texas, Radiation Dosimetry, Three-Dimensional
INGHAM, E., University of Leeds, Leeds, United Kingdom,
Hip Joints, Artificial
ISIK, CAN, Syracuse University, Syracuse, New York, Blood
Pressure Measurement
JAMES, SUSAN P., Colorado State University, Fort Collins,
Colorado, Biomaterials: Polymers
JENSEN, WINNIE, Aalborg University, Aalborg, Denmark,
Electroneurography
JIN, CHUNMING, North Carolina State University, Raleigh,
North Carolina, Biomaterials, Corrosion and Wear
of
JIN, Z.M., University of Leeds, Leeds, United Kingdom, Hip
Joints, Artificial
JOHNSON, ARTHUR T., University of Maryland College Park,
Maryland, Medical Engineering Societies and Organizations
JONES, JULIAN R., Imperial College London, London, United
Kingdom, Biomaterials: Bioceramics
JOSHI, ABHIJEET, Abbott Spine, Austin, Texas, Spinal
Implants
JUNG, RANU, Arizona State University, Tempe, Arizona,
Functional Electrical Stimulation
JURISSON, SILVIA S., University of Missouri Columbia,
Missouri, Radionuclide Production and Radioactive
Decay
KAEDING, PATRICIA J., Godfrey & Kahn S.C., Madison,
Wisconsin, Codes and Regulations: Medical Devices
KAMATH, CELIA C., Mayo Clinic, Rochester, Minnesota,
KANE, MOLLIE, Madison, Wisconsin, Contraceptive Devices
KATHERINE, ANDRIOLE P., Harvard Medical School, Boston,
Massachusetts, Picture Archiving and Communication
Systems
KATSAGGELOS, AGGELOS K., Northwestern University, Evanston, Illinois, DNA Sequencing
KATZ, J. LAWRENCE, University of Missouri-Kansas City,

Kansas City, Missouri, Bone and Teeth, Properties of
KESAVAN, SUNIL, Akebono Corporation, Farmington Hills,
Michigan, Linear Variable Differential Transformers
KHANG, GILSON, Chonbuk National University, Biomaterials:
Tissue Engineering and Scaffolds
KHAODHIAR, LALITA, Harvard Medical School, Boston, Massachusetts, Cutaneous Blood Flow, Doppler Measurement
of
KIM, MOON SUK, Korea Research Institutes of Chemical
Technology, Biomaterials: Tissue Engineering and Scaffolds
KIM, YOUNG KON, Inje University, Kimhae City, Korea,
Alloys, Shape Memory
KINDWALL, ERIC P., St. Lukes Medical Center, Milwaukee,
Wisconsin, Hyperbaric Oxygenation
KING, MICHAEL A., University of Massachusetts, North Worcester, Massachusetts, Nuclear Medicine, Computers in
KLEBE, ROBERT J., University of Texas, San Antonio, Texas,
Fluorescence Measurements
KLEIN, BURTON, Burton Klein Associates, Newton, Massachusetts, Gas and Vacuum Systems, Centrally Piped
Medical
KNOPER, STEVEN R., University of Arizona College of Medicine, Ventilatory Monitoring
KONTAXAKIS, GEORGE, Universidad Politecnica de Madrid,
Madrid, Spain, Positron Emission Tomography
KOTTKE-MARCHANT, KANDICE, The Cleveland Clinic Foundation, Cleveland, Ohio, Vascular Graft Prosthesis
KRIPFGANS, OLIVER, University of Michigan, Ann Arbor,
Michigan, Ultrasonic Imaging
KULKARNI, AMOL D., University of WisconsinMadison,
Madison, Wisconsin, Ocular Fundus Reflectometry,
Visual Field Testing
KUMARADAS, J. CARL, Ryerson University, Toronto, Ontario,
Canada, Hyperthermia, Interstitial
KUNICKA, JOLANTA, Bayer HealthCare LLC, Tarrytown, New
York, Differential Counts, Automated
KWAK, KWANJ JOO, University of Miami Miller School of
Medicine, Miami, Florida, Microscopy, Scanning Force
LAKES, RODERIC, University of Wisconsin-Madison, Bone
and Teeth, Properties of
LAKKIREDDY, DHANUNJAYA, The Cleveland Clinic Foundation, Cleveland, Ohio, Hyperthermia, Ultrasonic
LARSEN, COBY, Case Western Reserve University, Cleveland,
Ohio, Vascular Graft Prosthesis
LASTER, BRENDA H., Ben Gurion University of the Negev,
Beer Sheva, Israel, Monoclonal Antibodies
LATTA, LOREN, University of Miami, Coral Gables, Florida,
Rehabilitation, Orthotics in
LEDER, RON S., Universidad Nacional Autonoma de Mexico
Mexico, Distrito Federal, Continuous Positive Airway
Pressure
LEE, CHIN, Harvard Medical School, Boston, Massachusetts,
Radiotherapy Treatment Planning, Optimization of;
Hyperthermia, Interstitial
LEE, HAI BANG, Korea Research Institutes of Chemical
Technology, Biomaterials: Tissue Engineering and
Scaffolds
LEE, SANG JIN, Korea Research Institutes of Chemical
Technology, Biomaterials: Tissue Engineering and
Scaffolds
LEI, LIU, Department of General Engineering, Urbana,
Illinois, Bioinformatics
CONTRIBUTOR LIST
LEI, XING, Stanford University, Stanford, California, Radiation Dose Planning, Computer-Aided
LEWIS, MATTHEW C., Medical College of Wisconsin, Milwaukee, Wisconsin, Hyperbaric Oxygenation
LI, CHAODI, University of Notre Dame, Notre Dame, Indiana,
Bone Cement, Acrylic
LI, JONATHAN G., University of Florida, Gainesville, Florida,
Radiation Dose Planning, Computer-Aided
LI, QIAO, University of Michigan, Ann Arbor, Michigan,
Immunotherapy
LI, YANBIN, University of Arkansas, Fayetteville, Arkansas,
Piezoelectric Sensors
LIBOFF, A.R., Oakland University, Rochester, Michigan,
Bone Ununited Fracture and Spinal Fusion, Electrical
Treatment of
LIGAS, JAMES, University of Connecticut, Farmington, Connecticut, Respiratory Mechanics and Gas Exchange
LIMOGE, AIME, The Rene Descartes University of Paris, Paris,
France, Electroanalgesia, Systemic
LIN, PEI-JAN PAUL, Beth Israel Deaconess Medical Center,
Boston, Massachusets, Mammography
LIN, ZHIYUE, University of Kansas Medical Center, Kansas
City, Kansas, Electrogastrogram
LINEAWEAVER, WILLIAM C., Unive rsity of Mississippi Medical Center, Jackson, Mississippi, Hyperbaric Medicine
LIPPING, TARMO, Tampere University of Technology, Pori,
Finland, Monitoring in Anesthesia
LIU, XIAOHUA, The University of Michigan, Ann Arbor,
Michigan, Polymeric Materials
LLOYD, J.J., Regional Medical Physics Department, Newcastle-upon-Tyne, United Kingdom, Ultraviolet Radiation
in Medicine
LOEB, ROBERT, University of Arizona, Tuscon, Arizona,
Anesthesia Machines
LOPES DE MELO, PEDRO, State University of Rio de Janeiro,
Terreo Salas, Maracana, Thermistors
LOUDON, ROBERT G., Lung Sounds
LOW, DANIEL A., Washington University School of Medicine,
St. Louis, Missouri, Radiation Therapy Simulator
LU, LICHUN, Mayo Clinic, College of Medicine, Rochester,
Minnesota, Microscopy, Electron
LU, ZHENG FENG, Columbia University, New York, New
York, Screen-Film Systems
LYON, ANDREW W., University of Calgary, Calgary, Canada,
Flame Atomic Emission Spectrometry and Atomic
Absorption Spectrometry
LYON, MARTHA E., University of Calgary, Calgary, Canada,
Flame Atomic Emission Spectrometry and Atomic
Absorption Spectrometry
MA, C-M CHARLIE, Fox Chase Cancer Center, Philadelphia,
Pennsylvania, X-Ray Therapy Equipment, Low and Medium Energy
MACIA, NARCISO F., Arizona State University at the Polytechnic Campus, Mesa, Arizona, Pneumotachometers
MACKENZIE, COLIN F., University of Maryland, School of
Medicine, Shock, Treatment of
MACKIE, THOMAS R., University of Wisconsin, Madison,
Wisconsin, Tomotherapy
MADNANI, ANJU, LSU Medical Centre, Shreveport, Louisiana, Transcutaneous Electrical Nerve Stimulation
(TENS)
MADNANI, SANJAY, LSU Medical Centre, Shreveport, Louisiana, Transcutaneous Electrical Nerve Stimulation
(TENS)
ix
MADSEN, MARK T., University of Iowa, Iowa City, Iowa,

Anger Camera
MAGNANO, MAURO, ENT Division Mauriziano Hospital,
Torino, Italy, Drug Delivery Systems
MANDEL, RICHARD, Boston University School of Medicine,
Boston, Massachusetts, Colorimetry
MANNING, KEEFE B., Pennsylvania State University, University Park, Pennsylvania, Flowmeters
MAO, JEREMY J., University of Illinois at Chicago, Chicago,
Illinois, Cartilage and Meniscus, Properties of
MARCOLONGO, MICHELE, Drexel University, Philadelphia,
Pennsylvania, Spinal Implants
MAREK, MIROSLAV, Georgia Institute of Technology, Atlanta,
Georgia, Biomaterials, Corrosion and Wear of
MARION, NICHOLAS W., University of Illinois at Chicago,
Chicago, Illinois, Cartilage and Meniscus, Properties of
MASTERS, KRISTYN S., University of Wisconsin, Madison,
Wisconsin, Tissue Engineering
MAUGHAN, RICHARD L., Hospital of the University of Pennsylvania, Neutron Beam Therapy
MCADAMS, ERIC, University of Ulster at Jordanstown, Newtownabbey, Ireland, Bioelectrodes
MCARTHUR, SALLY L., University of Sheffield, Sheffield,
United Kingdom, Biomaterials, Surface Properties of
MCEWEN, MALCOM, National Research Council of Canada,
Ontario, Canada, Radiation Dosimetry for Oncology
MCGOWAN, EDWARD J., E.J. McGowan & Associates, Biofeedback
MCGRATH, SUSAN, Dartmouth College, Hanover, New Hampshire, Oxygen Analyzers
MEEKS, SANFORD L., University of Florida, Gainesville,
Florida, Radiosurgery, Stereotactic
MELISSA, PETER, University of Central Florida NanoScience
Technology Center, Orlando, Florida, Biosurface Engineering
MENDELSON, YITZHAK, Worcester Polytechnic Institute,
Optical Sensors
METZKER, MICHAEL L., Baylor College of Medicine, Houston,
Texas, Polymerase Chain Reaction
MEYEREND, M.E., University of WisconsinMadison,
Madison, Wisconsin, Magnetic Resonance Imaging
MICHLER, ROBERT, Montefiore Medical Center, Bronx, New
York, HeartLung Machines
MICIC, MIODRAG, MP Biomedicals LLC, Irvine, California,
Microscopy and Spectroscopy, Near-Field
MILLER, WILLIAM, University of Missouri Columbia,
Missouri, Radionuclide Production and Radioactive
Decay
MITTRA, ERIK, Stony Brook University, New York, Bone
Density Measurement
MODELL, MARK, Harvard Medical School, Boston, Massachusetts, Fiber Optics in Medicine
MORE, ROBERT B., RBMore Associates, Austin, Texas Biomaterials Carbon
MORE, ROBERT, Austin, Texas, Heart Valves, Prosthetic
MORROW, DARREN, Royal Adelaide Hospital, Adelaide,
Australia, Intraaortic Balloon Pump
MOURTADA, FIRAS, MD Anderson Cancer Center, Houston,
Texas, Brachytherapy, Intravascular
MOY, VINCENT T., University of Miami, Miller School of
Medicine, Miami, Florida, Microscopy, Scanning Force
MUFTU, SINAN, Northeastern University, Boston, Massachusetts, Tooth and Jaw, Biomechanics of
MURPHY, RAYMOND L.H., Lung Sounds
CONTRIBUTOR LIST
MURPHY, WILLIAM L., University of Wisconsin, Madison,

Wisconsin, Tissue Engineering
MURRAY, ALAN, Newcastle University Medical Physics, Newcastle upon Tyne, United Kingdom, Pace makers
MUTIC, SASA, Washington University School of Medicine, St.
Louis, Missouri, Radiation Therapy Simulator
NARAYAN, ROGER J., University of North Carolina, Chapel
Hill, North Carolina, Biomaterials, Corrosion and Wear of
NATALE, ANDREA, The Cleveland Clinic Foundation,
Cleveland, Ohio, Hyperthermia, Ultrasonic
NAZERAN, HOMER, The University of Texas, El Paso, Texas,
Electrocardiography, Computers in
NEUMAN, MICHAEL R., Michigan Technological University,
Houghton, Houghton, Michigan, Fetal Monitoring, Neonatal Monitoring
NEUZIL, PAVEL, Institute of Bioengineering and Nanotechnology, Singapore, Immunologically Sensitive FieldEffect Transistors
NICKOLOFF, EDWARD L., Columbia University, New York,
New York, X-Ray Quality Control Program
NIEZGODA, JEFFREY A., Medical College of Wisconsin,
Milwaukee, Wisconsin, Hyperbaric Oxygenation
NISHIKAWA, ROBERT M., The University of Chicago,
Chicago, Illinois, Computer-Assisted Detection and Diagnosis
NUTTER, BRIAN, Texas Tech University, Lubbock, Texas,
Medical Records, Computers in
ODONOHUE, WILLIAM, University of Nevada, Reno, Nevada,
Sexual Instrumentation
ORTON, COLIN, Harper Hospital and Wayne State University,
Detroit, Michigan, Medical Physics Literature
OZCELIK, SELAHATTIN, Texas A&M University, Kingsville,
Texas, Drug Infusion Systems
PANITCH, ALYSSA, Arizona State University, Tempe, Arizona,
Biomaterials: An Overview
PAOLINO, DONATELLA, University of Catanzaro Magna
Grcia, Germaneto (CZ), Italy, Drug Delivery Systems
PAPAIOANNOU, GEORGE, University of Wisconsin, Milwaukee,
Wisconsin, Joints, Biomechanics of
PARK, GRACE E., Purdue University, West Lafayette, Indiana, Porous Materials for Biological Applications
PARMENTER, BRETT A., State University of New York at
Buffalo, Buffalo, New York, Sleep Studies, Computer
Analysis of
PATEL, DIMPI, The Cleveland Clinic Foundation, Cleveland,
Ohio, Hyperthermia, Ultrasonic
PEARCE, JOHN, The University of Texas, Austin, Texas,
Electrosurgical Unit (ESU)
PELET, SERGE, Massachusetts Institute of Technology, Cambridge, Massachusetts, Microscopy, Fluorescence
PERIASAMY, AMMASI, University of Virginia, Charlottesville,
Virginia, Cellular Imaging
PERSONS, BARBARA L., University of Mississippi Medical
Center, Jackson, Mississippi, Hyperbaric Medicine
PIPER, IAN, The University of Memphis, Memphis,
Tennessee, Monitoring, Intracranial Pressure
POLETTO, CHRISTOPHER J., National Institutes of Health,
Tactile Stimulation
PREMINGER, GLENN M., Duke University Medical Center,
Durham, North Carolina, Lithotripsy
PRENDERGAST, PATRICK J., Trinity College, Dublin, Ireland,
Orthopedics, Prosthesis Fixation for
PREVITE, MICHAEL, Massachusetts Institute of Technology,
Cambridge, Massachusetts, Microscopy, Fluorescence
PURDY, JAMES A., UC Davis Medical Center, Sacramento,

California, Radiotherapy Accessories
QI, HAIRONG, Advanced Concepts Analysis, Inc., Falls
Church, Virginia, Thermography
QIN, YIXIAN, Stony Brook University, New York, Bone Density Measurement
QUAN, STUART F., University of Arizona, Tucson, Arizona,
Ventilatory Monitoring
QUIROGA, RODRIGO QUIAN, University of Leicester, Leicester,
United Kingdom, Evoked Potentials
RAHAGHI, FARBOD N., University of California, La Jolla,
California, Glucose Sensors
RAHKO, PETER S., University of Wisconsin Medical School,
Echocardiography and Doppler Echocardiography
RALPH, LIETO, University of WisconsinMadison, Madison,
Wisconsin, Codes and Regulations: Radiation
RAMANATHAN, LAKSHMI, Mount Sinai Medical Center, Analytical Methods, Automated
RAO, SATISH S.C., University of Iowa College of Medicine,
Iowa City, Iowa, Anorectal Manometry
RAPOPORT, DAVID M., NYU School of Medicine, New York,
New York, Continuous Positive Airway Pressure
REBELLO, KEITH J., The Johns Hopkins University Applied
Physics Lab, Laurel, Maryland, Micro surgery
REDDY, NARENDER, The University of Akron, Akron, Ohio,
Linear Variable Differential Transformers
REN-DIH, SHEU, Memorial Sloan-Kettering Cancer Center,
New York, New York, Radiation Therapy Treatment Planning, Monte Carlo Calculations in
RENGACHARY, SETTI S., Detroit, Michigan, Human Spine,
Biomechanics of
REPPERGER, DANIEL W., Wright-Patterson Air Force Base,
Dayton, Ohio, Human Factors in Medical Devices
RITCHEY, ERIC R., The Ohio State University, Columbus,
Ohio, Contact Lenses
RIVARD, MARK J., Tufts New England Medical Center, Boston, Massachusetts, Imaging Devices
ROBERTSON, J. DAVID, University of Missouri, Columbia,
Missouri, Radionuclide Production and Radioactive Decay
ROTH, BRADLEY J., Oakland University, Rochester, Michigan, Defibrillators
ROWE-HORWEGE, R. WANDA, University of Texas Medical
School, Houston, Texas, Hyperthermia, Systemic
RUMSEY, JOHN W., University of Central Florida, Orlando,
Florida, Biosurface Engineering
RUTKOWSKI, GREGORY E., University of Minnesota, Duluth,
Minnesota, Engineered Tissue
SALATA, O.V., University of Oxford, Oxford, United Kingdom, Nanoparticles
SAMARAS, THEODOROS, Aristotle University of Thessaloniki
Department of Physics, Thessaloniki, Greece, Thermometry
SANGOLE, ARCHANA P., Transitional Learning Center at
Galveston, Galveston, Texas, Rehabilitation, Computers
in Cognitive
SARKOZI, LASZLO, Mount Sinai School of Medicine, Analytical Methods, Automated
SCHEK, HENRY III, University of Michigan, Ann Arbor,
Michigan, Optical Tweezers
SCHMITZ, CHRISTOPH H., State University of New York Downstate Medical Center, Brooklyn, New York, Peripheral
Vascular Noninvasive Measurements
SCHUCKERS, STEPHANIE A.C., Clarkson University, Potsdam,
New York, Arrhythmia Analysis, Automated
CONTRIBUTOR LIST
SCOPE, KENNETH, Northwestern University, Chicago,

Illinois, Ventilators, Acute Medical Care
SCOTT, ADZICK N., University of Pennsylvania, Philadelphia,
Pennsylvania, Intrauterine Surgical Techniques
SEAL, BRANDON L., Arizona State University, Tempe,
Arizona, Biomaterials: An Overview
SEALE,
GARY,
Transitional
Learning
Center
at
in Cognitive
SEGERS, PATRICK, Ghent University, Belgium, Hemodynamics
SELIM, MOSTAFA A., Cleveland Metropolitan General Hospital, Palm Coast, Florida, Colposcopy
SETHI, ANIL, Loyola University Medical Center, Maywood,
Illinois, X-Rays: Interaction with Matter
SEVERINGHAUS, JOHN W., University of California in San
Francisco, CO2 Electrodes
SHALODI, ABDELWAHAB D., Cleveland Metropolitan General
Hospital, Palm Coast, Florida, Colposcopy
SHANMUGASUNDARAM, SHOBANA, New Jersey Institute of Technology, Newark, New Jersey, Polymeric Materials
SHARD, ALEXANDER G., University of Sheffield, Sheffield
United Kingdom, Biomaterials, Surface Properties of
SHEN, LI-JIUAN, National Taiwan University School of Pharmacy, Taipei, Taiwan, Colorimetry
SHEN, WEI-CHIANG,University of Southern California School
of Pharmacy, Los Angeles, California, Colorimetry
SHERAR, MICHAEL D., London Health Sciences Centre and
University of Western Ontario, London, Ontario, Canada,
Hyperthermia, Interstitial
SHERMAN, DAVID, The Johns Hopkins University, Baltimore,
Maryland, Electroencephalography
SHI, DONGLU, University of Cincinnati, Cincinnati, Ohio,
Biomaterials, Testing and Structural Properties of
SHUCARD, DAVID W.M., State University of New York at
Buffalo, Buffalo, New York, Sleep Studies, Computer
Analysis of
SIEDBAND, MELVIN P., University of Wisconsin, Madison,
Wisconsin, Image Intensifiers and Fluoroscopy
SILBERMAN, HOWARD, University of Southern California, Los
Angeles, California, Nutrition, Parenteral
SILVERMAN, GORDON, Manhattan College, Computers in the
Biomedical Laboratory
SILVERN, DAVID A., Medical Physics Unit, Rabin Medical
Center, Petah Tikva, Israel, Prostate Seed Implants
SINHA, PIYUSH, The Ohio State University, Columbus, Ohio,
Drug Delivery Systems
SINHA, ABHIJIT ROY, University of Cincinnati, Cincinnati, Ohio, Coronary Angioplasty and Guidewire
Diagnostics
SINKJR, THOMAS, Aalborg University, Aalborg, Denmark,
Electroneurography
SLOAN, JEFFREY A., Mayo Clinic, Rochester, Minnesota,
SO, PETER T.C., Massachusetts Institute of Technology,
Cambridge, Massachusetts, Microscopy, Fluorescence
SOBOL, WLAD T., University of Alabama at Birmingham
Health System, Birmingham, Alabama, Nuclear Magnetic Resonance Spectroscopy
SOOD, SANDEEP, University of Illinois at Chicago, Chicago,
SPECTOR, MYRON, Brigham and Womens Hospital, Boston,
Massachusetts, Biocompatibility of Materials
xi
SPELMAN, FRANCIS A., University of Washington, Cochlear

Prostheses
SRINIVASAN, YESHWANTH, Texas Tech University, Lubbock,
Texas, Medical Records, Computers in
SRIRAM, NEELAMEGHAM, University of Buffalo, Buffalo, New
York, Cell Counters, Blood
STARKO, KENTON R., Point Roberts, Washington, Physiological Systems Modeling
STARKSCHALL, GEORGE, The University of Texas, Radiotherapy, Three-Dimensional Conformal
STAVREV, PAVEL, Cross Cancer Institute, Edmonton, Alberta,
Canada, Radiotherapy Treatment Planning, Optimization of
STENKEN, JULIE A., Rensselaer Polytechnic Institute, Troy,
New York, Microdialysis Sampling
STIEFEL, ROBERT, University of Maryland Medical Center,
Baltimore, Maryland, Equipment Acquisition
STOKES, I.A.F., Polytechniquie Montreal, Montreal Quebec,
Canada, Scoliosis, Biomechanics of
STONE, M.H., University of Leeds, Leeds, United Kingdom,
SU, XIAo-LI, BioDetection Instruments LLC, Fayetteville,
Arkansas, Piezoelectric Sensors
SUBHAN, ARIF, Masterplan Technology Management,
Chatsworth, California, Equipment Maintenance,
Biomedical
SWEENEY, JAMES D., Arizona State University, Tempe,
Arizona, Functional Electrical Stimulation
SZETO, ANDREW Y.J., San Diego State University, San Diego,
California, Blind and Visually Impaired, Assistive Technology for
TAKAYAMA, SHUICHI, University of Michigan, Ann Arbor,
Michigan, Microbioreactors
TAMUL, PAUL C., Northwestern University, Chicago, Illinois,
Ventilators, Acute Medical Care
TAMURA, TOSHIYO, Chiba University School of Engineering,
Chiba, Japan, Home Health Care Devices
TANG, XIANGYANG, GE Healthcare Technologies, Wankesha,
Wisconsin, Computed Tomography Simulators
TAYLOR, B.C., The University of Akron, Akron, Ohio, Cardiac
Output, Indicator Dilution Measurement of
TEMPLE, RICHARD O., Transitional Learning Center at
in Cognitive
TEN, STANLEY, Salt Lake City, Utah, Electroanalgesia, Systemic
TERRY, TERESA M., Walter Reed Army Institute of
Research, Rockville, Maryland, Blood Collection and
Processing
THAKOR, N.V., Johns Hopkins University, Baltimore, Maryland, Neurological Monitors
THIERENS, HUBERT M.A., University of Ghent, Ghent, Belgium, Radiopharmaceutical Dosimetry
THOMADSEN, BRUCE, University of WisconsinMadison,
Madison, Wisconsin, Codes and Regulations: Radiation
TIPPER, J.L., University of Leeds, Leeds, United Kingdom,
TOGAWA, TATSUO, Waseda University, Saitama, Japan, Integrated Circuit Temperature Sensor
TORNAI, MARTIN, Duke University, Durham, North Carolina,
X-Ray Equipment Design
TRAN-SON-TAY, ROGER, University of Florida, Gainesville,
Florida, Blood Rheology
xii
CONTRIBUTOR LIST
TRAUTMAN, EDWIN D., RMF Strategies, Cambridge, Massachusetts, Cardiac Output, Thermodilution Measurement
of
TREENA, LIVINGSTON ARINZEH, New Jersey Institute of Technology, Newark, New Jersey, Polymeric Materials
TRENTMAN, TERRENCE L., Mayo Clinic Scottsdale, Spinal
Cord Stimulation
TROKEN, ALEXANDER J., University of Illinois at Chicago,
Chicago, Illinois, Cartilage and Meniscus, Properties of
TSAFTARIS, SOTIRIOS A., Northwestern University, Evanston,
Illinois, DNA Sequence
TSOUKALAS, D., NTUA, Athens, Attiki, Greece, Capacitive
Microsensors for Biomedical Applications
TULIPAN, NOEL, Vanderbilt University Medical Center,
Nashville, Tennessee, Intrauterine Surgical Techniques
TUTEJA, ASHOK K., University of Utah, Salt Lake City, Utah,
Anorectal Manometry
TY, SMITH N., University of California, San Diego, California, Physiological Systems Modeling
TYRER, HARRY W., University of Missouri-Columbia, Columbia, Missouri, Cytology, Automated
VALVANO, JONATHAN W., The University of Texas, Austin,
Texas, Bioheat Transfer
VAN DEN HEUVAL, FRANK, Wayne State University, Detroit,
Michigan, Imaging Devices
VEIT, SCHNABEL, Aalborg University, Aalborg, Denmark,
Electroneurography
VELANOVICH, VIC, Henry Ford Hospital, Detroit, Michigan,
Esophageal Manometry
VENKATASUBRAMANIAN,
GANAPRIYA,
Arizona
State
University, Tempe, Arizona, Functional Electrical
Stimulation
VERAART, CLAUDE, Catholique University of Louvain, Brussels, Belgium, Visual Prostheses
VERDONCK, PASCAL, Ghent University, Belgium, Hemodynamics
VERMARIEN, HERMAN, Vrije Universiteit Brussel, Brussels,
Belgium, Phonocardiography, Recorders, Graphic
VEVES, ARISTIDIS, Harvard Medical School, Boston, Massachusetts, Cutaneous Blood Flow, Doppler Measurement
of
VICINI, PAOLO, University of Washington, Seattle, Washington, Pharmacokinetics and Pharmacodynamics
VILLE, JA NTTI, Tampere University of Technology, Pori,
VRBA, JINI, VSM MedTech Ltd., Biomagnetism
WAGNER, THOMAS, H., M. D. Anderson Cancer Center
Orlando, Orlando, Florida, Radiosurgery, Stereotactic
WAHLEN, GEORGE E., Veterans Affairs Medical Center and
the University of Utah, Salt Lake City, Utah, Anorectal
Manometry
WALKER, GLENN M., North Carolina State University,
Raleigh, North Carolina, Microfluidics
WALTERSPACHER, DIRK, The Johns Hopkins University, Baltimore, Maryland, Electroencephalography
WAN, LEO Q., Liu Ping, Columbia University, New York,
New York, Cartilage and Meniscus, Properties of
WANG, GE, University of Iowa, Iowa City, Iowa, Computed
Tomography Simulators
WANG, HAIBO, Louisiana State University Health Center
Shreveport, Louisiana, Monitoring, Umbilical Artery
and Vein, Ambulatory Monitoring
WANG, HONG, Wayne State University, Detroit, Michigan,
Anesthesia, Computers in
WANG, LE YI, Wayne State University, Detroit, Michigan,

Anesthesia, Computers in
WANG, QIAN, A & M University Health Science Center,
Dallas, Texas, Strain Gages
WARWICK, WARREN J., University of Minnesota Medical
School, Minneapolis, Minnesota, Cystic Fibrosis Sweat
Test
WATANABE, YOICHI, Columbia University Radiation
Oncology, New York, New York, Phantom Materials in
Radiology
WAXLER, MORRIS, Godfrey & Kahn S.C., Madison, Wisconsin, Codes and Regulations: Medical Devices
WEBSTER, THOMAS J., Purdue University, West Lafayette,
Indiana, Porous Materials for Biological Applications
WEGENER, JOACHIM, University of Oslo, Oslo, Norway, Impedance Spectroscopy
WEI, SHYY, University of Michigan, Ann Arbor, Michigan,
Blood Rheology
WEINMEISTER, KENT P., Mayo Clinic Scottsdale, Spinal Cord
Stimulation
WEIZER, ALON Z., Duke University Medical Center, Durham,
North Carolina, Lithotripsy
WELLER, PETER, City University , London, United Kingdom,
Intraaortic Balloon Pump
WELLS, JASON, LSU Medical Centre, Shreveport, Louisiana,
Transcutaneous Electrical Nerve Stimulation (TENS)
WENDELKEN, SUZANNE, Dartmouth College, Hanover, New
Hampshire, Oxygen Analyzers
WHELAN, HARRY T., Medical College of Wisconsin, Milwaukee, Wisconsin, Hyperbaric Oxygenation
WHITE, ROBERT, Memorial Hospital, Regional Newborn
Program, South Bend, Indiana, Incubators, Infant
WILLIAMS, LAWRENCE E., City of Hope, Duarte, California,
Nuclear Medicine Instrumentation
WILSON, KERRY, University of Central Florida, Orlando,
Florida, Biosurface Engineering
WINEGARDEN, NEIL, University Health Network Microarray
Centre, Toronto, Ontario, Canada, Microarrays
WOJCIKIEWICZ, EWA P., University of Miami Miller School
of Medicine, Miami, Florida, Microscopy, Scanning
Force
WOLBARST, ANTHONY B., Georgetown Medical School,
Washington, DC, Radiotherapy Treatment Planning,
Optimization of
WOLF, ERIK, University of Pittsburgh, Pittsburgh, Pennsylvania, Mobility Aids
WOOD, ANDREW, Swinburne University of Technology, Melbourne, Australia, Nonionizing Radiation, Biological
Effects of
WOODCOCK, BRIAN, University of Michigan, Ann Arbor,
Michigan, Blood, Artificial
WREN, JOAKIM, Linkoping University, Linkoping, Sweden,
Thermocouples
XIANG, ZHOU, Brigham and Womens Hospital, Boston, Massachusetts, Biocompatibility of Materials
XUEJUN, WEN, Clemson University, Clemson, South
Carolina, Biomaterials, Testing and Structural Properties of
YAN, ZHOU, University of Notre Dame, Notre Dame, Indiana,
Bone Cement, Acrylic
YANNAS, IOANNIS V., Massachusetts Institute of Technology,
Skin Tissue Engineering for Regeneration
YASZEMSKI, MICHAEL J., Mayo Clinic, College of Medicine,
Rochester, Minnesota, Microscopy, Electron
CONTRIBUTOR LIST
YENI, YENER N., Henry Ford Hospital, Detroit, Michigan,

Joints, Biomechanics of
YLI-HANKALA, ARVI, Tampere University of Technology, Pori,
YOKO, KAMOTANI, University of Michigan, Ann Arbor, Michigan, Microbioreactors
YOON, KANG JI, Korea Institute of Science and Technology,
Seoul, Korea, Micropower for Medical Applications
YORKE, ELLEN, Memorial Sloan-Kettering Cancer Center,
New York, New York, Radiation Therapy Treatment Planning, Monte Carlo Calculations in
YOSHIDA, KEN, Aalborg University, Aalborg, Denmark, Electroneurography
YOUNGSTEDT, SHAWN D., University of South Carolina,
Columbia, South Carolina, Sleep Laboratory
YU, YIH-CHOUNG, Lafayette College, Easton, Pennsylvania,
Blood Pressure, Automatic Control of
ZACHARIAH, EMMANUEL S., University of Medicine and Dentistry of New Jersey, New Brunswick, New Jersey, Immunologically Sensitive Field-Effect Transistors
xiii
ZAIDER, MARCO, Memorial Sloan Kettering Cancer Center,

New York, New York, Prostate Seed Implants
ZAPANTA, CONRAD M., Penn State College of Medicine,
Hershey, Pennsylvania, Heart, Artificial
ZARDENETA, GUSTAVO, University of Texas, San Antonio,
Texas, Fluorescence Measurements
ZELMANOVIC, DAVID, Bayer HealthCare LLC, Tarrytown,
New York, Differential Counts, Automated
ZHANG, MIN, University of Washington, Seattle, Washington,
Biomaterials: Polymers
ZHANG, YI, University of Buffalo, Buffalo, New York, Cell
Counters, Blood
ZHU, XIAOYUE, University of Michigan, Ann Arbor, Michigan, Microbioreactors
ZIAIE, BABAK, Purdue University, W. Lafayette, Indiana,
Biotelemetry
ZIELINSKI, TODD M., Medtronic, Inc., Minneapolis, Minnesota, Bioimpedance in Cardiovascular Medicine
ZIESSMAN, HARVEY A., Johns Hopkins University, Computed
Tomography, Single Photon Emission
PREFACE
The Encyclopedia of Medical Devices and Instrumentation is excellent for browsing and searching for those new
divergent associations that may advance work in a peripheral field. While it can be used as a reference for facts, the
articles are long enough that they can serve as an educational instrument and provide genuine understanding of a
subject.
One can use this work just as one would use a dictionary,
since the articles are arranged alphabetically by topic. Cross
references assist the reader looking for subjects listed under
slightly different names. The index at the end leads the
reader to all articles containing pertinent information on
any subject. Listed on pages xxi to xxx are all the abbreviations and acronyms used in the Encyclopedia. Because of
the increasing use of SI units in all branches of science, these
units are provided throughout the Encyclopedia articles as
well as on pages xxxi to xxxv in the section on conversion
factors and unit symbols.
I owe a great debt to the many people who have contributed to the creation of this work. At John Wiley & Sons,
Encyclopedia Editor George Telecki provided the idea and
guiding influence to launch the project. Sean Pidgeon was
Editorial Director of the project. Assistant Editors Roseann
Zappia, Sarah Harrington, and Surlan Murrell handled the
myriad details of communication between publisher, editor,
authors, and reviewers and stimulated authors and
reviewers to meet necessary deadlines.
My own background has been in the electrical aspects of
biomedical engineering. I was delighted to have the assistance of the editorial board to develop a comprehensive
encyclopedia. David J. Beebe suggested cellular topics such
as microfluidics. Jerry M. Calkins assisted in defining the
chemically related subjects, such as anesthesiology.
Michael R. Neuman suggested subjects related to sensors,
such as in his own workneonatology. Joon B. Park has
written extensively on biomaterials and suggested related
subjects. Edward S. Sternick provided many suggestions
from medical physics. The Editorial Board was instrumental both in defining the list of subjects and in suggesting
authors.
This second edition brings the field up to date. It is
available on the web at http://www.mrw.interscience.wiley.
com/emdi, where articles can be searched simultaneously to
provide rapid and comprehensive information on all aspects
of medical devices and instrumentation.
This six-volume work is an alphabetically organized compilation of almost 300 articles that describe critical aspects of
medical devices and instrumentation.
It is comprehensive. The articles emphasize the contributions of engineering, physics, and computers to each of the
general areas of anesthesiology, biomaterials, burns, cardiology, clinical chemistry, clinical engineering, communicative disorders, computers in medicine, critical care
medicine, dermatology, dentistry, ear, nose, and throat,
emergency medicine, endocrinology, gastroenterology,
genetics, geriatrics, gynecology, hematology, heptology,
internal medicine, medical physics, microbiology, nephrology, neurology, nutrition, obstetrics, oncology, ophthalmology, orthopedics, pain, pediatrics, peripheral vascular
disease, pharmacology, physical therapy, psychiatry, pulmonary medicine, radiology, rehabilitation, surgery, tissue
engineering, transducers, and urology.
The discipline is defined through the synthesis of the core
knowledge from all the fields encompassed by the application of engineering, physics, and computers to problems in
medicine. The articles focus not only on what is now useful
but also on what is likely to be useful in future medical
applications.
These volumes answer the question, What are the
branches of medicine and how does technology assist each
of them? rather than What are the branches of technology
and how could each be used in medicine? To keep this work
to a manageable length, the practice of medicine that is
unassisted by devices, such as the use of drugs to treat
disease, has been excluded.
The articles are accessible to the user; each benefits from
brevity of condensation instead of what could easily have
been a book-length work. The articles are designed not for
peers, but rather for workers from related fields who wish to
take a first look at what is important in the subject.
The articles are readable. They do not presume a detailed
background in the subject, but are designed for any person
with a scientific background and an interest in technology.
Rather than attempting to teach the basics of physiology or
Ohms law, the articles build on such basic concepts to show
how the worlds of life science and physical science meld to
produce improved systems. While the ideal reader might be
a person with a Masters degree in biomedical engineering or
medical physics or an M.D. with a physical science undergraduate degree, much of the material will be of value to
others with an interest in this growing field. High school
students and hospital patients can skip over more technical
areas and still gain much from the descriptive presentations.
JOHN G. WEBSTER
University of Wisconsin, Madison
xv
LIST OF ARTICLES
CARDIAC OUTPUT, FICK TECHNIQUE FOR
CARDIAC OUTPUT, INDICATOR DILUTION
MEASUREMENT OF
CARDIAC OUTPUT, THERMODILUTION
MEASUREMENT OF
CARDIOPULMONARY RESUSCITATION
CARTILAGE AND MENISCUS, PROPERTIES OF
CELL COUNTERS, BLOOD
CELLULAR IMAGING
CHROMATOGRAPHY
CO2 ELECTRODES
COBALT 60 UNITS FOR RADIOTHERAPY
COCHLEAR PROSTHESES
CODES AND REGULATIONS: MEDICAL DEVICES
CODES AND REGULATIONS: RADIATION
COLORIMETRY
COLPOSCOPY
COMMUNICATION DEVICES
COMMUNICATIVE DISORDERS, COMPUTER
APPLICATIONS FOR
COMPUTED TOMOGRAPHY
COMPUTED TOMOGRAPHY SCREENING
COMPUTED TOMOGRAPHY SIMULATORS
COMPUTED TOMOGRAPHY, SINGLE PHOTON
EMISSION
COMPUTER-ASSISTED DETECTION AND DIAGNOSIS
COMPUTERS IN THE BIOMEDICAL LABORATORY
CONTACT LENSES
CONTINUOUS POSITIVE AIRWAY PRESSURE
CONTRACEPTIVE DEVICES
CORONARY ANGIOPLASTY AND GUIDEWIRE
DIAGNOSTICS
CRYOSURGERY
CUTANEOUS BLOOD FLOW, DOPPLER
MEASUREMENT OF
CYSTIC FIBROSIS SWEAT TEST
CYTOLOGY, AUTOMATED
DEFIBRILLATORS
DIFFERENTIAL COUNTS, AUTOMATED
DIGITAL ANGIOGRAPHY
DNA SEQUENCE
DRUG DELIVERY SYSTEMS
DRUG INFUSION SYSTEMS
ECHOCARDIOGRAPHY AND DOPPLER
ECHOCARDIOGRAPHY
ELECTROANALGESIA, SYSTEMIC
ELECTROCARDIOGRAPHY, COMPUTERS IN
ELECTROCONVULSIVE THERAPY
ELECTROENCEPHALOGRAPHY
ELECTROGASTROGRAM
ELECTROMYOGRAPHY
ELECTRONEUROGRAPHY
ELECTROPHORESIS
ALLOYS, SHAPE MEMORY

AMBULATORY MONITORING
ANALYTICAL METHODS, AUTOMATED
ANESTHESIA MACHINES
ANESTHESIA, COMPUTERS IN
ANGER CAMERA
ANORECTAL MANOMETRY
ARRHYTHMIA ANALYSIS, AUTOMATED
ARTERIES, ELASTIC PROPERTIES OF
AUDIOMETRY
BIOCOMPATIBILITY OF MATERIALS
BIOELECTRODES
BIOFEEDBACK
BIOHEAT TRANSFER
BIOIMPEDANCE IN CARDIOVASCULAR MEDICINE
BIOINFORMATICS
BIOMAGNETISM
BIOMATERIALS, ABSORBABLE
BIOMATERIALS: AN OVERVIEW
BIOMATERIALS: BIOCERAMICS
BIOMATERIALS: CARBON
BIOMATERIALS, CORROSION AND WEAR OF
BIOMATERIALS FOR DENTISTRY
BIOMATERIALS: POLYMERS
BIOMATERIALS, SURFACE PROPERTIES OF
BIOMATERIALS, TESTING AND STRUCTURAL
PROPERTIES OF
BIOMATERIALS: TISSUE ENGINEERING AND
SCAFFOLDS
BIOMECHANICS OF EXERCISE FITNESS
BIOMEDICAL ENGINEERING EDUCATION
BIOSURFACE ENGINEERING
BIOTELEMETRY
BLADDER DYSFUNCTION, NEUROSTIMULATION
OF
BLIND AND VISUALLY IMPAIRED, ASSISTIVE
TECHNOLOGY FOR
BLOOD COLLECTION AND PROCESSING
BLOOD GAS MEASUREMENTS
BLOOD PRESSURE MEASUREMENT
BLOOD PRESSURE, AUTOMATIC CONTROL OF
BLOOD RHEOLOGY
BLOOD, ARTIFICIAL
BONE AND TEETH, PROPERTIES OF
BONE CEMENT, ACRYLIC
BONE DENSITY MEASUREMENT
BONE UNUNITED FRACTURE AND SPINAL FUSION,
ELECTRICAL TREATMENT OF
BORON NEUTRON CAPTURE THERAPY
BRACHYTHERAPY, HIGH DOSAGE RATE
BRACHYTHERAPY, INTRAVASCULAR
CAPACITIVE MICROSENSORS FOR BIOMEDICAL
APPLICATIONS
xvii
xviii
LIST OF ARTICLES
ELECTROPHYSIOLOGY
ELECTRORETINOGRAPHY
ELECTROSURGICAL UNIT (ESU)
ENDOSCOPES
ENGINEERED TISSUE
ENVIRONMENTAL CONTROL
EQUIPMENT ACQUISITION
EQUIPMENT MAINTENANCE, BIOMEDICAL
ESOPHAGEAL MANOMETRY
EVOKED POTENTIALS
EXERCISE STRESS TESTING
EYE MOVEMENT, MEASUREMENT TECHNIQUES FOR
FETAL MONITORING
FIBER OPTICS IN MEDICINE
FLAME ATOMIC EMISSION SPECTROMETRY AND
ATOMIC ABSORPTION SPECTROMETRY
FLOWMETERS
FLUORESCENCE MEASUREMENTS
FUNCTIONAL ELECTRICAL STIMULATION
GAMMA KNIFE
GAS AND VACUUM SYSTEMS, CENTRALLY PIPED
MEDICAL
GASTROINTESTINAL HEMORRHAGE
GLUCOSE SENSORS
HEART VALVE PROSTHESES
HEART VALVE PROSTHESES, IN VITRO FLOW
DYNAMICS OF
HEART VALVES, PROSTHETIC
HEART, ARTIFICIAL
HEARTLUNG MACHINES
HEAT AND COLD, THERAPEUTIC
HEMODYNAMICS
HIGH FREQUENCY VENTILATION
HIP JOINTS, ARTIFICIAL
HOME HEALTH CARE DEVICES
HUMAN FACTORS IN MEDICAL DEVICES
HUMAN SPINE, BIOMECHANICS OF
HYDROCEPHALUS, TOOLS FOR DIAGNOSIS
AND TREATMENT OF
HYPERBARIC MEDICINE
HYPERBARIC OXYGENATION
HYPERTHERMIA, INTERSTITIAL
HYPERTHERMIA, SYSTEMIC
HYPERTHERMIA, ULTRASONIC
IMAGE INTENSIFIERS AND FLUOROSCOPY
IMAGING DEVICES
IMMUNOLOGICALLY SENSITIVE FIELD-EFFECT
TRANSISTORS
IMMUNOTHERAPY
IMPEDANCE PLETHYSMOGRAPHY
IMPEDANCE SPECTROSCOPY
INCUBATORS, INFANT
INTEGRATED CIRCUIT TEMPERATURE SENSOR
INTRAAORTIC BALLOON PUMP
INTRAUTERINE SURGICAL TECHNIQUES
IONIZING RADIATION, BIOLOGICAL EFFECTS OF
ION-SENSITIVE FIELD-EFFECT TRANSISTORS
JOINTS, BIOMECHANICS OF
LARYNGEAL PROSTHETIC DEVICES
LENSES, INTRAOCULAR
LIGAMENT AND TENDON, PROPERTIES OF
LINEAR VARIABLE DIFFERENTIAL TRANSFORMERS

LITHOTRIPSY
LIVER TRANSPLANTATION
LUNG SOUNDS
MAGNETIC RESONANCE IMAGING
MAMMOGRAPHY
MEDICAL EDUCATION, COMPUTERS IN
MEDICAL ENGINEERING SOCIETIES
AND ORGANIZATIONS
MEDICAL GAS ANALYZERS
MEDICAL PHYSICS LITERATURE
MEDICAL RECORDS, COMPUTERS IN
MICROARRAYS
MICROBIAL DETECTION SYSTEMS
MICROBIOREACTORS
MICRODIALYSIS SAMPLING
MICROFLUIDICS
MICROPOWER FOR MEDICAL APPLICATIONS
MICROSCOPY AND SPECTROSCOPY, NEAR-FIELD
MICROSCOPY, CONFOCAL
MICROSCOPY, ELECTRON
MICROSCOPY, FLUORESCENCE
MICROSCOPY, SCANNING FORCE
MICROSCOPY, SCANNING TUNNELING
MICROSURGERY
MINIMALLY INVASIVE SURGERY
MOBILITY AIDS
MONITORING IN ANESTHESIA
MONITORING, HEMODYNAMIC
MONITORING, INTRACRANIAL PRESSURE
MONITORING, UMBILICAL ARTERY AND VEIN
MONOCLONAL ANTIBODIES
NANOPARTICLES
NEONATAL MONITORING
NEUROLOGICAL MONITORS
NEUTRON ACTIVATION ANALYSIS
NEUTRON BEAM THERAPY
NONIONIZING RADIATION, BIOLOGICAL EFFECTS OF
NUCLEAR MAGNETIC RESONANCE SPECTROSCOPY
NUCLEAR MEDICINE INSTRUMENTATION
NUCLEAR MEDICINE, COMPUTERS IN
NUTRITION, PARENTERAL
OCULAR FUNDUS REFLECTOMETRY
OCULAR MOTILITY RECORDING AND NYSTAGMUS
OFFICE AUTOMATION SYSTEMS
OPTICAL SENSORS
OPTICAL TWEEZERS
ORTHOPEDIC DEVICES, MATERIALS AND
DESIGN FOR
ORTHOPEDICS, PROSTHESIS FIXATION FOR
OXYGEN ANALYZERS
OXYGEN MONITORING
PACEMAKERS
PANCREAS, ARTIFICIAL
PERIPHERAL VASCULAR NONINVASIVE
MEASUREMENTS
PHANTOM MATERIALS IN RADIOLOGY
PHARMACOKINETICS AND PHARMACODYNAMICS
PHONOCARDIOGRAPHY
PHOTOGRAPHY, MEDICAL
PHYSIOLOGICAL SYSTEMS MODELING
LIST OF ARTICLES
PICTURE ARCHIVING AND COMMUNICATION

SYSTEMS
PIEZOELECTRIC SENSORS
PNEUMOTACHOMETERS
POLYMERASE CHAIN REACTION
POLYMERIC MATERIALS
POROUS MATERIALS FOR BIOLOGICAL
APPLICATIONS
POSITRON EMISSION TOMOGRAPHY
PROSTATE SEED IMPLANTS
PULMONARY PHYSIOLOGY
QUALITY-OF-LIFE MEASURES, CLINICAL
SIGNIFICANCE OF
RADIATION DOSE PLANNING, COMPUTER-AIDED
RADIATION DOSIMETRY FOR ONCOLOGY
RADIATION DOSIMETRY, THREE-DIMENSIONAL
RADIATION PROTECTION INSTRUMENTATION
RADIATION THERAPY, INTENSITY MODULATED
RADIATION THERAPY SIMULATOR
RADIATION THERAPY TREATMENT PLANNING,
MONTE CARLO CALCULATIONS IN
RADIATION THERAPY, QUALITY ASSURANCE IN
RADIOLOGY INFORMATION SYSTEMS
RADIONUCLIDE PRODUCTION AND
RADIOACTIVE DECAY
RADIOPHARMACEUTICAL DOSIMETRY
RADIOSURGERY, STEREOTACTIC
RADIOTHERAPY ACCESSORIES
RADIOTHERAPY, HEAVY ION
RADIOTHERAPY, INTRAOPERATIVE
RADIOTHERAPY, THREE-DIMENSIONAL CONFORMAL
RADIOTHERAPY TREATMENT PLANNING,
OPTIMIZATION OF
RECORDERS, GRAPHIC
REHABILITATION AND MUSCLE TESTING
REHABILITATION, COMPUTERS IN COGNITIVE
REHABILITATION, ORTHOTICS IN
RESIN-BASED COMPOSITES
RESPIRATORY MECHANICS AND GAS EXCHANGE
SAFETY PROGRAM, HOSPITAL
SCOLIOSIS, BIOMECHANICS OF
SCREEN-FILM SYSTEMS
xix
SEXUAL INSTRUMENTATION
SHOCK, TREATMENT OF
SKIN SUBSTITUTE FOR BURNS, BIOACTIVE
SKIN TISSUE ENGINEERING FOR REGENERATION
SKIN, BIOMECHANICS OF
SLEEP LABORATORY
SLEEP STUDIES, COMPUTER ANALYSIS OF
SPINAL CORD STIMULATION
SPINAL IMPLANTS
STATISTICAL METHODS
STEREOTACTIC SURGERY
STERILIZATION OF BIOLOGIC SCAFFOLD
MATERIALS
STRAIN GAGES
TACTILE STIMULATION
TELERADIOLOGY
TEMPERATURE MONITORING
THERMISTORS
THERMOCOUPLES
THERMOGRAPHY
THERMOMETRY
TISSUE ABLATION
TISSUE ENGINEERING
TOMOTHERAPY
TONOMETRY, ARTERIAL
TOOTH AND JAW, BIOMECHANICS OF
TRACER KINETICS
TRANSCUTANEOUS ELECTRICAL NERVE
STIMULATION (TENS)
ULTRASONIC IMAGING
ULTRAVIOLET RADIATION IN MEDICINE
VASCULAR GRAFT PROSTHESIS
VENTILATORS, ACUTE MEDICAL CARE
VENTILATORY MONITORING
VISUAL FIELD TESTING
VISUAL PROSTHESES
X-RAY EQUIPMENT DESIGN
X-RAY QUALITY CONTROL PROGRAM
X-RAY THERAPY EQUIPMENT, LOW AND
MEDIUM ENERGY
X-RAYS: INTERACTION WITH MATTER
X-RAYS, PRODUCTION OF
ABBREVIATIONS AND ACRONYMS

AAMI
AAPM
ABC
ABET
ABG
ABLB
ABS
ac
AC
ACA
ACES
ACL
ACLS
ACOG
ACR
ACS
A/D
ADC
ADCC
ADCL
ADP
A-D-T
AE
AEA
AEB
AEC
AED
AEMB
AES
AESC
AET
AFO
AGC
AHA
AI
AICD
AID
AIDS
AL
ALG
ALS
Association for the Advancement of

Medical Instrumentation
American Association of Physicists in
Medicine
Automatic brightness control
Accreditation board for engineering
training
Arterial blood gases
Alternative binaural loudness balance
Acrylonitrilebutadienestyrene
Alternating current
Abdominal circumference; Affinity
chromatography
Automated clinical analyzer
Augmentative communication evaluation
system
Anterior chamber lens
Advanced cardiac life support
American College of Obstetrics and
Gynecology
American College of Radiology
American Cancer Society; American
College of Surgeons
Analog-to-digital
Agar diffusion chambers; Analog-todigital converter
Antibody-dependent cellular
cytotoxicity
Accredited Dosimetry Calibration
Laboratories
Adenosine diphosphate
Admission, discharge, and transfer
Anion exchange; Auxiliary electrode
Articulation error analysis
Activation energy barrier
Automatic exposure control
Automatic external defibrillator
Alliance for Engineering in Medicine
and Biology
Auger electron spectroscopy
American Engineering Standards
Committee
Automatic exposure termination
Ankle-foot orthosis
Automatic gain control
American Heart Association
Arterial insufficiency
Automatic implantable cardiac
defibrillator
Agency for International Development
Acquired immune deficiency syndrome
Anterior leaflet
Antilymphocyte globulin
ALT
ALU
AM
AMA
amu
ANOVA
ANSI
AP
APD
APL
APR
AR
Ara-C
ARD
ARDS
ARGUS
ARMA
ARMAX
AS
ASA
ASCII
ASD
ASHE
ASTM
AT
ATA
ATLS
ATN
ATP
ATPD
ATPS
ATR
AUC
AUMC
AV
AZT
BA
BAEP
BAPN
BAS
BASO
BB
BBT
xxi
Advanced life support; Amyotropic

lateral sclerosis
Alanine aminotransferase
Arithmetic and logic unit
Amplitude modulation
American Medical Association
Atomic mass units
Analysis of variance
American National Standards Institute
Action potential; Alternative pathway;
Anteroposterior
Anterioposterior diameter
Adjustable pressure limiting valve;
Applied Physics Laboratory
Anatomically programmed radiography
Amplitude reduction; Aortic
regurgitation; Autoregressive
Arabinosylcytosine
Absorption rate density
Adult respiratory distress syndrome
Arrhythmia guard system
Autoregressive-moving-average model
Autoregressive-moving-average model
with external inputs
Aortic stenosis
American Standards Association
American standard code for information
interchange
Antisiphon device
American Society for Hospital Engineering
American Society for Testing and
Materials
Adenosine-thiamide; Anaerobic threshold;
Antithrombin
Atmosphere absolute
Advanced trauma life support
Acute tubular necrosis
Adenosine triphosphate
Ambient temperature pressure dry
Ambient temperature pressure
saturated
Attenuated total reflection
Area under curve
Area under moment curve
Atrioventricular
Azido thymidine
Biliary atresia
Brainstem auditory evoked potential
Beta-amino-proprionitryl
Boston anesthesis system
Basophil
Buffer base
Basal body temperature
xxii
BCC
BCD
BCG
BCLS
BCRU
BDI
BE
BET
BH
BI
BIH
BIPM
BJT
BMDP
BME
BMET
BMO
BMR
BOL
BP
BR
BRM
BRS
BSS
BTG
BTPS
BUN
BW
CA
CABG
CAD/CAM
CAD/D
CADD
CAI
CAM
cAMP
CAPD
CAPP
CAT
CATS
CAVH
CB
CBC
CBF
CBM
CBV
CC
CCC
CCD
CCE
CCF
CCL
CCM
CCPD
Body-centered cubic
Binary-coded decimal
Ballistocardiogram
Basic cardiac life support
British Commitee on Radiation Units
and Measurements
Beck depression inventory
Base excess; Binding energy
Brunauer, Emmett, and Teller methods
His bundle
Biological indicators
Beth Israel Hospital
International Bureau of Weights and
Measurements
Bipolar junction transistor
Biomedical Programs
Biomedical engineering
Biomedical equipment technician
Biomechanically optimized
Basal metabolic rate
Beginning of life
Bereitschafts potential; Break point
Polybutadiene
Biological response modifier
Bibliographic retrieval services
Balanced salt solution
Beta thromboglobulin
Body temperature pressure saturated
Blood urea nitrogen
Body weight
Conductive adhesives
Coronary artery by-pass grafting
Computer-aided design/computer-aided
manufacturing
Computer-aided drafting and design
Central axis depth dose
Computer assisted instruction;
Computer-aided instruction
Computer-assisted management
Cyclic AMP
Continuous ambulatory peritoneal
dialysis
Child amputee prosthetic project
Computerized axial tomography
Computer-assisted teaching system;
Computerized aphasia treatment system
Continuous arteriovenous hemofiltration
Conjugated bilirubin; Coulomb barrier
Complete blood count
Cerebral blood flow
Computer-based management
Cerebral blood volume
Closing capacity
Computer Curriculum Company
Charge-coupled device
Capacitance contact electrode
Cross-correlation function
Cardiac catheterization laboratory
Critical care medical services
Continuous cycling peritoneal
dialysis
CCTV
CCU
CD
CDR
CDRH
CEA
CF
CFC
CFR
CFU
CGA
CGPM
CHO
CHO
CI
CICU
CIF
CIN
CK
CLAV
CLSA
CM
CMAD
CMI
CMRR
CMV
CNS
CNV
CO
COBAS
COPD
COR
CP
CPB
CPET
CPM
CPP
CPR
cps
CPU
CR
CRBB
CRD
CRL
CRT
CS
CSA
CSF
CSI
CSM
CT
CTI
CV
Closed circuit television system

Coronary care unit; Critical care unit
Current density
Complimentary determining region
Center for Devices and Radiological
Health
Carcinoembryonic antigen
Conversion factor; Cystic fibrosis
Continuous flow cytometer
Code of Federal Regulations
Colony forming units
Compressed Gas Association
General Conference on Weights and
Measures
Carbohydrate
Chinese hamster ovary
Combination index
Cardiac intensive care unit
Contrast improvement factor
Cervical intraepithelial neoplasia
Creatine kinase
Clavicle
Computerized language sample analysis
Cardiomyopathy; Code modulation
Computer managed articulation diagnosis
Computer-managed instruction
Common mode rejection ratio
Conventional mechanical ventilation;
Cytomegalovirus
Central nervous system
Contingent negative variation
Carbon monoxide; Cardiac output
Comprehensive Bio-Analysis System
Chronic obstructive pulmonary disease
Center of rotation
Cerebral palsy; Closing pressure; Creatine
phosphate
Cardiopulmonary bypass
Cardiac pacemaker electrode tips
Computerized probe measurements
Cerebral perfusion pressure;
Cryoprecipitated plasma
Cardiopulmonary resuscitation
Cycles per second
Central Processing unit
Center of resistance; Conditioned response;
Conductive rubber; Creatinine
Complete right bundle branch block
Completely randomized design
Crown rump length
Cathode ray tube
Conditioned stimulus; Contrast scale;
Crown seat
Compressed spectral array
Cerebrospinal fluid
Chemical shift imaging
Chemically sensitive membrane
Computed tomography; Computerized
tomography
Cumulative toxicity response index
Closing volume
C.V.
CVA
CVP
CVR
CW
CWE
CWRU
DAC
DAS
dB
DB
DBMS
DBS
dc
DCCT
DCP
DCS
DDC
DDS
DE
DEN
DERS
DES
d.f.
DHCP
DHE
DHEW
DHHS
DHT
DI
DIC
DIS
DL
DLI
DM
DME
DN
DNA
DOF
DOS
DOT-NHTSA
DPB
DPG
DQE
DRESS
DRG
DSA
DSAR
DSB
DSC
D-T
DTA
d.u.
DUR
DVT
EA
EB
EBCDIC
Coefficient of variation
Cerebral vascular accident
Central venous pressure
Cardiovascular resistance
Continuous wave
Coated wire electrodes
Case Western Reserve University
Digital-to-analog converter
Data acquisition system
Decibel
Direct body
Data base management system
Deep brain stimulation
Direct current
Diabetes control and complications trial
Distal cavity pressure
Dorsal column stimulation
Deck decompression chamber
Deep diving system
Dispersive electrode
Device experience network
Drug exception ordering system
Diffuse esophageal spasm
Distribution function
Distributed Hospital Computer Program
Dihematoporphyrin ether
Department of Health Education and
Welfare
Department of Health and Human Services
Duration of hypothermia
Deionized water
Displacement current
Diagnostic interview schedule
Double layer
Difference lumen for intensity
Delta modulation
Dropping mercury electrode
Donation number
Deoxyribonucleic acid
Degree of freedom
Drug ordering system
Department of Transportation Highway
Traffic Safety Administration
Differential pencil beam
Diphosphoglycerate
Detection quantum efficiency
Depth-resolved surface coil spectroscopy
Diagnosis-related group
Digital subtraction angiography
Differential scatter-air ratio
Double strand breaks
Differential scanning calorimetry
Deuterium-on-tritium
Differential thermal analysis
Density unit
Duration
Deep venous thrombosis
Esophageal accelerogram
Electron beam
Extended binary code decimal interchange
code
EBS
EBV
EC
ECC
ECCE
ECD
ECG
ECM
ECMO
ECOD
ECRI
ECS
ECT
EDD
EDP
EDTA
EDX
EEG
EEI
EELV
EER
EF
EF
EFA
EGF
EGG
EIA
EIU
ELF
ELGON
ELISA
ELS
ELV
EM
EMBS
emf
EMG
EMGE
EMI
EMS
EMT
ENT
EO
EOG
EOL
EOS
EP
EPA
ER
ERCP
ERG
ERMF
ERP
ERV
xxiii
Early burn scar

EpsteinBarr Virus
Ethyl cellulose
Emergency cardiac care; Extracorporeal
circulation
Extracapsular cataract extinction
Electron capture detector
Electrocardiogram
Electrochemical machining
Extracorporeal membrane oxygenation
Extracranial cerebrovascular occlusive
disease
Emergency Care Research Institute
Exners Comprehensive System
Electroconvulsive shock therapy;
Electroconvulsive therapy; Emission
computed tomography
Estimated date of delivery
Aortic end diastolic pressure
Ethylenediaminetetraacetic acid
Energy dispersive X-ray analysis
Electroencephalogram
Electrode electrolyte interface
End-expiratory lung volume
Electrically evoked response
Ejection fraction
Electric field; Evoked magnetic fields
Estimated fetal age
Epidermal growth factor
Electrogastrogram
Enzyme immunoassay
Electrode impedance unbalance
Extra low frequency
Electrical goniometer
Enzyme-linked immunosorbent assay
Energy loss spectroscopy
Equivalent lung volume
Electromagnetic
Engineering in Medicine and Biology
Society
Electromotive force
Electromyogram
Integrated electromyogram
Electromagnetic interference
Emergency medical services
Emergency medical technician
Ear, nose, and throat
Elbow orthosis
Electrooculography
End of life
Eosinophil
Elastoplastic; Evoked potentiate
Environmental protection agency
Evoked response
Endoscopic retrograde
cholangiopancreatography
Electron radiography;
Electroretinogram
Event-related magnetic field
Event-related potential
Expiratory reserve volume
xxiv
ESCA
ESI
ESRD
esu
ESU
ESWL
ETO, Eto
ETT
EVA
EVR
EW
FAD
FARA
FBD
FBS
fcc
FCC
Fct
FDA
FDCA
FE
FECG
FEF
FEL
FEM
FEP
FES
FET
FEV
FFD
FFT
FGF
FHR
FIC
FID
FIFO
FITC
FL
FM
FNS
FO
FO-CRT
FP
FPA
FR
FRC
FSD
FTD
FTIR
FTMS
FU
FUDR
FVC
FWHM
FWTM
GABA
GAG
GBE
Electron spectroscopy for chemical

analysis
Electrode skin impedance
End-stage renal disease
Electrostatic unit
Electrosurgical unit
Extracorporeal shock wave lithotripsy
Ethylene oxide
Exercise tolerance testing
Ethylene vinyl acetate
Endocardial viability ratio
Extended wear
Flavin adenine dinucleotide
Flexible automation random analysis
Fetal biparietal diameter
Fetal bovine serum
Face centered cubic
Federal Communications Commission
Fluorocrit
Food and Drug Administration
Food, Drug, and Cosmetic Act
Finite element
Fetal electrocardiogram
Forced expiratory flow
Free electron lasers
Finite element method
Fluorinated ethylene propylene
Functional electrical stimulation
Field-effect transistor
Forced expiratory volume
Focal spot to film distance
Fast Fourier transform
Fresh gas flow
Fetal heart rate
Forced inspiratory capacity
Flame ionization detector; Free-induction
decay
First-in-first-out
Fluorescent indicator tagged polymer
Femur length
Frequency modulation
Functional neuromuscular stimulation
Foramen ovale
Fiber optics cathode ray tube
Fluorescence polarization
Fibrinopeptide A
Federal Register
Federal Radiation Council; Functional
residual capacity
Focus-to-surface distance
Focal spot to tissue-plane distance
Fourier transform infrared
Fourier transform mass spectrometer
Fluorouracil
Floxuridine
Forced vital capacity
Full width at half maximum
Full width at tenth maximum
Gamma amino buteric acid
Glycosaminoglycan
Gas-bearing electrodynamometer
GC
GDT
GFR
GHb
GI
GLC
GMV
GNP
GPC
GPH
GPH-EW
GPO
GSC
GSR
GSWD
HA
HAM
Hb
HBE
HBO
HC
HCA
HCFA
HCL
hcp
HCP
HDPE
HECS
HEMS
HEPA
HES
HETP
HF
HFCWO
HFER
HFJV
HFO
HFOV
HFPPV
HFV
HHS
HIBC
HIMA
HIP
HIS
HK
HL
HMBA
HMO
HMWPE
HOL
HP
HpD
HPLC
HPNS
HPS
HPX
Gas chromatography; Guanine-cytosine

Gas discharge tube
Glomerular filtration rate
Glycosylated hemoglobin
Gastrointestinal
Gasliquid chromatography
General minimum variance
Gross national product
Giant papillary conjunctivitis
Gas-permeable hard
Gas-permeable hard lens extended wear
Government Printing Office
Gas-solid chromatography
Galvanic skin response
Generalized spike-wave discharge
Hydroxyapatite
Helical axis of motion
Hemoglobin
His bundle electrogram
Hyperbaric oxygenation
Head circumference
Hypothermic circulatory arrest
Health care financing administration
Harvard Cyclotron Laboratory
Hexagonal close-packed
Half cell potential
High density polyethylene
Hospital Equipment Control System
Hospital Engineering Management
System
High efficiency particulate air filter
Hydroxyethylstarch
Height equivalent to a theoretical plate
High-frequency; Heating factor
High-frequency chest wall oscillation
High-frequency electromagnetic radiation
High-frequency jet ventilation
High-frequency oscillator
High-frequency oscillatory ventilation
High-frequency positive pressure
ventilation
High-frequency ventilation
Department of Health and Human
Services
Health industry bar code
Health Industry Manufacturers
Association
Hydrostatic indifference point
Hospital information system
Hexokinase
Hearing level
Hexamethylene bisacetamide
Health maintenance organization
High-molecular-weight polyethylene
Higher-order languages
Heating factor; His-Purkinje
Hematoporphyrin derivative
High-performance liquid chromatography
High-pressure neurological syndrome
His-Purkinje system
High peroxidase activity
HR
HRNB
H/S
HSA
HSG
HTCA
HTLV
HU
HVL
HVR
HVT
IA
IABP
IAEA
IAIMS
IASP
IC
ICCE
ICD
ICDA
ICL
ICP
ICPA
ICRP
ICRU
ICU
ID
IDDM
IDE
IDI
I:E
IEC
IEEE
IEP
BETS
IF
IFIP
IFMBE
IGFET
IgG
IgM
IHP
IHSS
II
IIIES
IM
IMFET
Heart rate; High-resolution

Halstead-Reitan Neuropsychological
Battery
Hard/soft
Human serum albumin
Hysterosalpingogram
Human tumor cloning assay
Human T cell lymphotrophic virus
Heat unit; Houndsfield units; Hydroxyurea
Half value layer
Hypoxic ventilatory response
Half-value thickness
Image intensifier assembly; Inominate
artery
Intraaortic balloon pumping
International Atomic Energy Agency
Integrated Academic Information
Management System
International Association for the Study
of Pain
Inspiratory capacity; Integrated circuit
Intracapsular cataract extraction
Intracervical device
International classification of diagnoses
Ms-clip lens
Inductively coupled plasma;
Intracranial pressure
Intracranial pressure amplitude
International Commission on
Radiological Protection
International Commission on Radiological
Units and Measurements
Intensive care unit
Inside diameter
Insulin dependent diabetes mellitus
Investigational device exemption
Index of inspired gas distribution
Inspiratory: expiratory
International Electrotechnical
Commission; Ion-exchange
chromatography
Institute of Electrical and Electronics
Engineers
Individual educational program
Inelastic electron tunneling spectroscopy
Immunofluorescent
International Federation for Information
Processing
International Federation for Medical and
Biological Engineering
Insulated-gate field-effect transistor
Immunoglobulin G
Immunoglobulin M
Inner Helmholtz plane
Idiopathic hypertrophic subaortic stenosis
Image intensifier
Image intensifier input-exposure
sensitivity
Intramuscular
Immunologically sensitive field-effect
transistor
IMIA
IMS
IMV
INF
IOL
IPC
IPD
IPG
IPI
IPPB
IPTS
IR
IRB
IRBBB
IRPA
IRRAS
IRRS
IRS
IRV
IS
ISC
ISDA
ISE
ISFET
ISIT
ISO
ISS
IT
ITEP
ITEPI
ITLC
IUD
IV
IVC
IVP
JCAH
JND
JRP
KB
Kerma
KO
KPM
KRPB
LA
LAD
LAE
LAK
LAL
LAN
LAP
LAT
LBBB
LC
xxv
International Medical Informatics

Association
Information management system
Intermittent mandatory ventilation
Interferon
Intraocular lens
Ion-pair chromatography
Intermittent peritoneal dialysis
Impedance plethysmography
Interpulse interval
Intermittent positive pressure breathing
International practical temperature scale
Polyisoprene rubber
Institutional Review Board
Incomplete right bundle branch block
International Radiation Protection
Association
Infrared reflection-absorption
spectroscopy
Infrared reflection spectroscopy
Internal reflection spectroscopy
Inspiratory reserve capacity
Image size; Ion-selective
Infant skin servo control
Instantaneous screw displacement axis
Ion-selective electrode
Ion-sensitive field effect transistor
Intensified silicon-intensified target tube
International Organization for
Standardization
Ion scattering spectroscopy
Intrathecal
Institute of Theoretical and Experimental
Physics
Instantaneous trailing edge pulse
impedance
Instant thin-layer chromatography
Intrauterine device
Intravenous
Inferior vena cava
Intraventricular pressure
Joint Commission on the Accreditation
of Hospitals
Just noticeable difference
Joint replacement prosthesis
Kent bundle
Kinetic energy released in unit mass
Knee orthosis
Kilopond meter
Krebs-Ringer physiological buffer
Left arm; Left atrium
Left anterior descending; Left axis
deviation
Left atrial enlargement
Lymphokine activated killer
Limulus amoebocyte lysate
Local area network
Left atrial pressure
Left anterior temporalis
Left bundle branch block
Left carotid; Liquid chromatography
xxvi
LCC
LCD
LDA
LDF
LDH
LDPE
LEBS
LED
LEED
LES
LESP
LET
LF
LH
LHT
LL
LLDPE
LLPC
LLW
LM
LNNB
LOS
LP
LPA
LPC
LPT
LPV
LRP
LS
LSC
LSI
LSV
LTI
LUC
LV
LVAD
LVDT
LVEP
LVET
LVH
LYMPH
MAA
MAC
MAN
MAP
MAST
MBA
MBV
MBX
MCA
MCG
MCI
MCMI
MCT
MCV
MDC
MDI
Left coronary cusp

Liquid crystal display
Laser Doppler anemometry
Laser Doppler flowmetry
Lactate dehydrogenase
Low density polyethylene
Low-energy brief stimulus
Light-emitting diode
Low energy electron diffraction
Lower esophageal sphincter
Lower esophageal sphincter pressure
Linear energy transfer
Low frequency
Luteinizing hormone
Local hyperthermia
Left leg
Linear low density polyethylene
Liquid-liquid partition chromatography
Low-level waste
Left masseter
Luria-Nebraska Neuropsychological
Battery
Length of stay
Late potential; Lumboperitoneal
Left pulmonary artery
Linear predictive coding
Left posterior temporalis
Left pulmonary veins
Late receptor potential
Left subclavian
Liquid-solid adsorption chromatography
Large scale integrated
Low-amplitude shear-wave
viscoelastometry
Low temperature isotropic
Large unstained cells
Left ventricle
Left ventricular assist device
Linear variable differential transformer
Left ventricular ejection period
Left ventricular ejection time
Left ventricular hypertrophy
Lymphocyte
Macroaggregated albumin
Minimal auditory capabilities
Manubrium
Mean airway pressure; Mean arterial
pressure
Military assistance to safety and traffic
Monoclonal antibody
Maximum breathing ventilation
Monitoring branch exchange
Methyl cryanoacrylate
Magnetocardiogram
Motion Control Incorporated
Millon Clinical Multiaxial Inventory
Microcatheter transducer
Mean corpuscular volume
Medical diagnostic categories
Diphenylmethane diisocyanate;
Medical Database Informatics
MDP
MDR
MDS
ME
MED
MEDPAR
MEFV
MEG
MeSH
METS
MF
MFP
MGH
MHV
MI
MIC
MIFR
MINET
MIR
MIS
MIT
MIT/BIH
MMA
MMA
MMECT
MMFR
mm Hg
MMPI
MMSE
MO
MONO
MOSFET
MP
MPD
MR
MRG
MRI
MRS
MRT
MS
MSR
MTBF
MTF
MTTR
MTX
MUA
MUAP
MUAPT
MUMPI
MUMPS
MV
MVO2
MVTR
MVV
MW
Mean diastolic aortic pressure

Medical device reporting
Multidimensional scaling
Myoelectric
Minimum erythema dose
Medicare provider analysis and review
Maximal expiratory flow volume
Magnetoencephalography
Medline subject heading
Metabolic equivalents
Melamine-formaldehyde
Magnetic field potential
Massachusetts General Hospital
Magnetic heart vector
Myocardial infarction
Minimum inhibitory concentration
Maximum inspiratory flow rate
Medical Information Network
Mercury-in-rubber
Medical information system;
Metal-insulator-semiconductor
Massachusetts Institute of Technology
Massachusetts Institute of Technology/
Beth Israel Hospital
Manual metal arc welding
Methyl methacrylate
Multiple-monitored ECT
Maximum midexpiratory flow rate
Millimeters of mercury
Minnesota Multiphasic Personality
Inventory
Minimum mean square error
Membrane oxygenation
Monocyte
Metal oxide silicon field-effect
transistor
Mercaptopurine; Metacarpal-phalangeal
Maximal permissible dose
Magnetic resonance
Magnetoretinogram
Magnetic resonance imaging
Magnetic resonance spectroscopy
Mean residence time
Mild steel; Multiple sclerosis
Magnetically shielded room
Mean time between failure
Modulation transfer function
Mean time to repair
Methotroxate
Motor unit activity
Motor unit action potential
Motor unit action potential train
Missouri University Multi-Plane
Imager
Massachusetts General Hospital utility
multiuser programming system
Mitral valve
Maximal oxygen uptake
Moisture vapor transmission rate
Maximum voluntary ventilation
Molecular weight
NAA
NAD
NADH
NADP
NAF
NARM
NBB
NBD
N-BPC
NBS
NCC
NCCLS
NCRP
NCT
NEEP
NEMA
NEMR
NEQ
NET
NEUT
NFPA
NH
NHE
NHLBI
NIR
NIRS
NK
NMJ
NMOS
NMR
NMS
NPH
NPL
NR
NRC
NRZ
NTC
NTIS
NVT
NYHA
ob/gyn
OCR
OCV
OD
ODC
ODT
ODU
OER
OFD
OHL
OHP
OIH
Neutron activation analysis

Nicotinamide adenine dinucleotide
Nicotinamide adenine dinucleotide,
reduced form
Nicotinamide adenine dinucleotide
phosphate
Neutrophil activating factor
Naturally occurring and acceleratorproduced radioactive materials
Normal buffer base
Neuromuscular blocking drugs
Normal bonded phase chromatography
National Bureau of Standards
Noncoronary cusp
National Committee for Clinical
Laboratory Standards; National
Committee on Clinical Laboratory
Standards
National Council on Radiation Protection
Neutron capture theory
Negative end-expiratory pressure
National Electrical Manufacturers
Association
Nonionizing electromagnetic radiation
Noise equivalent quanta
Norethisterone
Neutrophil
National Fire Protection Association
Neonatal hepatitis
Normal hydrogen electrode
National Heart, Lung, and Blood Institute
Nonionizing radiation
National Institute for Radiologic Science
Natural killer
Neuromuscular junction
N-type metal oxide silicon
Nuclear magnetic resonance
Neuromuscular stimulation
Normal pressure hydrocephalus
National Physical Laboratory
Natural rubber
Nuclear Regulatory Commission
Non-return-to-zero
Negative temperature coefficient
National Technical Information Service
Neutrons versus time
New York Heart Association
Obstetrics and gynecology
Off-center ratio; Optical character
recognition
Open circuit voltage
Optical density; Outside diameter
Oxyhemoglobin dissociation curve
Oxygen delivery truck
Optical density unit
Oxygen enhancement ratio
Object to film distance; Occiputo-frontal
diameter
Outer Helmholtz layer
Outer Helmholtz plane
Orthoiodohippurate
OPG
OR
OS
OTC
OV
PA
PACS
PAD
PAM
PAN
PAP
PAR
PARFR
PARR
PAS
PASG
PBI
PBL
PBT
PC
PCA
PCG
PCI
PCL
PCR
PCRC
PCS
PCT
PCWP
PD
PDD
PDE
p.d.f.
PDL
PDM
PDMSX
PDS
PE
PEEP
PEFR
PEN
PEP
PEPPER
PET
PEU
PF
PFA
PFC
PFT
PG
xxvii
Ocular pneumoplethysmography
Operating room
Object of known size; Operating system
Over the counter
Offset voltage
Posterioanterior; Pulmonary artery;
Pulse amplitude
Picture archiving and communications
systems
Primary afferent depolarization
Pulse amplitude modulation
Polyacrylonitrile
Pulmonary artery pressure
Photoactivation ratio
Program for Applied Research on
Fertility Regulation
Poetanesthesia recovery room
Photoacoustic spectroscopy
Pneumatic antishock garment
Penile brachial index
Positive beam limitation
Polybutylene terephthalate
Paper chromatography; Personal
computer; Polycarbonate
Patient controlled analgesia; Principal
components factor analysis
Phonocardiogram
Physiological cost index
Polycaprolactone; Posterior chamber
lens
Percent regurgitation
Perinatal Clinical Research Center
Patient care system
Porphyria cutanea tarda
Pulmonary capillary wedge pressure
Peritoneal dialysis; Poly-p-dioxanone;
Potential difference; Proportional and
derivative
Percent depth dose; Perinatal Data
Directory
Pregelled disposable electrodes
Probability density function
Periodontal ligament
Pulse duration modulation
Polydimethyl siloxane
Polydioxanone
Polyethylene
Positive end-expiratory pressure
Peak expiratory now rate
Parenteral and enteral nutrition
Preejection period
Programs examine phonetic find
phonological evaluation records
Polyethylene terephthalate;
Positron-emission tomography
Polyetherurethane
Platelet factor
Phosphonoformic add
Petrofluorochemical
Pulmonary function testing
Polyglycolide; Propylene glycol
xxviii
PGA
PHA
PHEMA
PI
PID
PIP
PL
PLA
PLATO
PLD
PLED
PLT
PM
PMA
p.m.f.
PMMA
PMOS
PMP
PMT
PO
Po2
POBT
POM
POMC
POPRAS
PP
PPA
PPF
PPM
PPSFH
PR
PRBS
PRP
PRO
PROM
PS
PSA
PSF
PSI
PSP
PSR
PSS
PT
PTB
PTC
PTCA
PTFE
PTT
PUL
Polyglycolic add
Phytohemagglutinin; Pulse-height
analyzer
Poly-2-hydroxyethyl methacrylate
Propidium iodide
Pelvic inflammatory disease;
Proportional/integral/derivative
Peak inspiratory pressure
Posterior leaflet
Polylactic acid
Program Logic for Automated Teaching
Operations
Potentially lethal damage
Periodic latoralized epileptiform discharge
Platelet
Papillary muscles; Preventive
maintenance
Polymethyl acrylate
Probability mass function
Polymethyl methacrylate
P-type metal oxide silicon
Patient management problem;
Poly(4-methylpentane)
Photomultiplier tube
Per os
Partial pressure of oxygen
Polyoxybutylene terephthalate
Polyoxymethylene
Patient order management and
communication system
Problem Oriented Perinatal Risk
Assessment System
Perfusion pressure; Polyproplyene;
Postprandial (after meals)
Phonemic process analysis
Plasma protein fraction
Pulse position modulation
Polymerized phyridoxalated stroma-free
hemoglobin
Pattern recognition; Pulse rate
Pseudo-random binary signals
Pulse repetition frequency
Professional review organization
Programmable read only memory
Polystyrene
Pressure-sensitive adhesive
Point spread function
Primary skin irritation
Postsynaptic potential
Proton spin resonance
Progressive systemic sclerosis
Plasma thromboplastin
Patellar tendon bearing orthosis
Plasma thromboplastin component;
Positive temperature coefficient;
Pressurized personal transfer capsule
Percutaneous transluminal coronary
angioplasty
Polytetrafluoroethylene
Partial thromboplastin time
Percutaneous ultrasonic lithotripsy
PURA
PUVA
P/V
PVC
PVI
PW
PWM
PXE
QA
QC
R-BPC
R/S
RA
RAD
RAE
RAM
RAP
RAT
RB
RBBB
RBC
RBE
RBF
RBI
RCBD
rCBF
RCC
RCE
R&D
r.e.
RE
REM
REMATE
RES
RESNA
RF
RFI
RFP
RFQ
RH
RHE
RIA
RM
RMR
RMS
RN
RNCA
ROI
ROM
RP
RPA
RPP
RPT
RPV
RQ
Prolonged ultraviolet-A radiation

Psoralens and longwave ultraviolet light
photochemotherapy
Pressure/volume
Polyvinyl chloride; Premature ventricular
contraction
Pressurevolume index
Pulse wave; Pulse width
Pulse width modulation
Pseudo-xanthoma elasticum
Quality assurance
Quality control
Reverse bonded phase chromatography
Radiopaque-spherical
Respiratory amplitude; Right arm
Right axis deviation
Right atrial enlargement
Random access memory
Right atrial pressure
Right anterior temporalis
Right bundle
Right bundle branch block
Red blood cell
Relative biologic effectiveness
Rose bengal fecal excretion
Resting baseline impedance
Randomized complete block diagram
Regional cerebral blood flow
Right coronary cusp
Resistive contact electrode
Research and development
Random experiment
Reference electrode
Rapid eye movement; Return electrode
monitor
Remote access and telecommunication
system
Reticuloendothelial system
Rehabilitation Engineering Society of
North America
Radio frequency; Radiographicnuoroscopic
Radio-frequency interference
Request for proposal
Request for quotation
Relative humidity
Reversible hydrogen electrode
Radioimmunoassay
Repetition maximum; Right masseter
Resting metabolic rate
Root mean square
Radionuclide
Radionuclide cineagiogram
Regions of interest
Range of motion; Read only memory
Retinitis pigmentosa
Right pulmonary artery
Rate pressure product
Rapid pull-through technique
Right pulmonary veins
Respiratory quotient
RR
RRT
RT
RTD
RTT
r.v.
RV
RVH
RVOT
RZ
SA
SACH
SAD
SAINT
SAL
SALT
SAMI
SAP
SAR
SARA
SBE
SBR
SC
SCAP
SCE
SCI
SCRAD
SCS
SCUBA
SD
SDA
SDS
S&E
SE
SEC
SEM
SEP
SEXAFS
SF
SFD
SFH
SFTR
SG
SGF
SGG
SGOT
SGP
SHE
SI
Recovery room
Recovery room time; Right posterior
temporalis
Reaction time
Resistance temperature device
Revised token test
Random variable
Residual volume; Right ventricle
Right ventricular hypertrophy
Right ventricular outflow tract
Return-to-zero
Sinoatrial; Specific absorption
Solid-ankle-cushion-heel
Source-axis distance; Statistical
Analysis System
System analysis of integrated network
of tasks
Sterility assurance level; Surface
averaged lead
Systematic analysis of language
transcripts
Socially acceptable monitoring
instrument
Systemic arterial pressure
Scatter-air ratio; Specific absorption rate
System for anesthetic and respiratory
gas analysis
Subbacterial endocarditis
Styrene-butadiene rubbers
Stratum corneum; Subcommittees
Right scapula
Saturated calomel electrode; Sister
chromatid exchange
Spinal cord injury
Sub-Committee on Radiation Dosimetry
Spinal cord stimulation
Self-contained underwater breathing
apparatus
Standard deviation
Stepwise discriminant analysis
Sodium dodecyl sulfate
Safety and effectiveness
Standard error
Size exclusion chromatography
Scanning electron microscope; Standard
error of the mean
Somatosensory evoked potential
Surface extended X-ray absorption
fine structure
Surviving fraction
Source-film distance
Stroma-free hemoglobin
Sagittal frontal transverse rotational
Silica gel
Silica gel fraction
Spark gap generator
Serum glutamic oxaloacetic transaminase
Strain gage plethysmography;
Stress-generated potential
Standard hydrogen electrode
Le Syste`me International dUnite s
SEBS
SID
SIMFU
SIMS
SISI
SL
SLD
SLE
SMA
SMAC
SMR
S/N
S:N/D
SNP
SNR
SOA
SOAP
SOBP
SP
SPECT
SPL
SPRINT
SPRT
SPSS
SQUID
SQV
SR
SRT
SS
SSB
SSD
SSE
SSEP
SSG
SSP
SSS
STD
STI
STP
STPD
SV
SVC
SW
TAA
TAC
TAD
TAG
TAH
TAR
TC
TCA
TCD
TCES
xxix
Surgical isolation barrier system

Source to image reception distance
Scanned intensity modulated focused
ultrasound
Secondary ion mass spectroscopy; System
for isometric muscle strength
Short increment sensitivity index
Surgical lithotomy
Sublethal damage
Systemic lupus erythemotodes
Sequential multiple analyzer
Sequential multiple analyzer with
computer
Sensorimotor
Signal-to-noise
Signal-to-noise ratio per unit dose
Sodium nitroprusside
Signal-to-noise ratio
Sources of artifact
Subjective, objective, assessment, plan
Spread-out Bragg peak
Skin potential
Single photon emission computed
tomography
Sound pressure level
Single photon ring tomograph
Standard platinum resistance
thermometer
Statistical Package for the Social Sciences
Superconducting quantum interference
device
Square wave voltammetry
Polysulfide rubbers
Speech reception threshold
Stainless steel
Single strand breaks
Source-to-skin distance; Source-to-surface
distance
Stainless steel electrode
Somatosensory evoked potential
Solid state generator
Skin stretch potential
Sick sinus syndrome
Source-tray distance
Systolic time intervals
Standard temperature and pressure
Standard temperature pressure dry
Stroke volume
Superior vena cava
Standing wave
Tumor-associated antigens
Time-averaged concentration
Transverse abdominal diameter
Technical Advisory Group
Total artificial heart
Tissue-air ratio
Technical Committees
Tricarboxylic acid cycle
Thermal conductivity detector
Transcutaneous cranial electrical
stimulation
xxx
TCP
TDD
TDM
TE
TEAM
TEM
TENS
TEP
TEPA
TF
TFE
TI
TICCIT
TLC
TLD
TMJ
TMR
TNF
TOF
TP
TPC
TPD
TPG
TPN
TR
tRNA
TSH
TSS
TTD
TTI
TTR
TTV
TTY
TUR
TURP
TV
TVER
TW
TxB2
TZ
UES
UP
UfflS
UHMW
Tricalcium phosphate
Telecommunication devices for the
deaf
Therapeutic drug monitoring
Test electrode; Thermoplastic elastomers
Technology evaluation and acquisition
methods
Transmission electron microscope;
Transverse electric and magnetic mode;
Transverse electromagnetic mode
Transcutaneous electrical nerve
stimulation
Tracheoesophageal puncture
Triethylenepho-sphoramide
Transmission factor
Tetrafluorethylene
Totally implantable
Time-shared Interaction ComputerControlled Information Television
Thin-layer chromatography; Total
lung capacity
Thermoluminescent dosimetry
Temporomandibular joint
Tissue maximum ratio; Topical
magnetic resonance
Tumor necrosis factor
Train-of-four
Thermal performance
Temperature pressure correction
Triphasic dissociation
Transvalvular pressure gradient
Total parenteral nutrition
Temperature rise
Transfer RNA
Thyroid stimulating hormone
Toxic shock syndrome
Telephone devices for the deaf
Tension time index
Transition temperature range
Trimming tip version
Teletypewriter
Transurethral resection
Transurethral resections of the
prostrate
Television; Tidal volume; Tricuspid valve
Transscleral visual evoked response
Traveling wave
Thrombozame B2
Transformation zone
Upper esophageal sphincter
Urea-formaldehyde
University Hospital Information System
Ultra high molecular weight
UHMWPE
UL
ULF
ULTI
UMN
UO
UPTD
UR
US
USNC
USP
UTS
UV
UVR
V/F
VA
VAS
VBA
VC
VCO
VDT
VECG
VEP
VF
VOP
VP
VPA
VPB
VPR
VSD
VSWR
VT
VTG
VTS
VV
WAIS-R
WAK
WAML
WBAR
WBC
WG
WHO
WLF
WMR
w/o
WORM
WPW
XPS
XR
YAG
ZPL
Ultra high molecular weight polyethylene

Underwriters Laboratory
Ultralow frequency
Ultralow temperature isotropic
Upper motor neuron
Urinary output
Unit pulmonary oxygen toxicity doses
Unconditioned response
Ultrasound; Unconditioned stimulus
United States National Committee
United States Pharmacopeia
Ultimate tensile strength
Ultraviolet; Umbilical vessel
Ultraviolet radiation
Voltage-to-frequency
Veterans Administration
Visual analog scale
Vaginal blood volume in arousal
Vital capacity
Voltage-controlled oscillator
Video display terminal
Vectorelectrocardiography
Visually evoked potential
Ventricular fibrillation
Venous occlusion plethysmography
Ventriculoperitoneal
Vaginal pressure pulse in arousal
Ventricular premature beat
Volume pressure response
Ventricular septal defect
Voltage standing wave ratio
Ventricular tachycardia
Vacuum tube generator
Viewscan text system
Variable version
Weschler Adult Intelligence
Scale-Revised
Wearable artificial kidney
Wide-angle mobility light
Whole-body autoradiography
White blood cell
Working Groups
World Health Organization; Wrist hand
orthosis
Williams-Landel-Ferry
Work metabolic rate
Weight percent
Write once, read many
Wolff-Parkinson-White
X-ray photon spectroscopy
Xeroradiograph
Yttrium aluminum garnet
Zero pressure level
CONVERSION FACTORS AND UNIT SYMBOLS

SI UNITS (ADOPTED 1960)
A new system of metric measurement, the International System of Units (abbreviated SI), is being implemented throughout
the world. This system is a modernized version of the MKSA (meter, kilogram, second, ampere) system, and its details are
published and controlled by an international treaty organization (The International Bureau of Weights and Measures).
SI units are divided into three classes:
Base Units
length
massz
time
electric current
thermodynamic temperature
amount of substance
luminous intensity
metery (m)
kilogram (kg)
second (s)
ampere (A)
kelvin (K)
mole (mol)
candela (cd)
Supplementary Units
plane angle
solid angle
radian (rad)
steradian (sr)
Derived Units and Other Acceptable Units

These units are formed by combining base units, supplementary units, and other derived units. Those derived units having
special names and symbols are marked with an asterisk (*) in the list below:
Quantity
*
absorbed dose
acceleration
*
activity (of ionizing radiation source)
area
Unit
gray
meter per second squared
becquerel
square kilometer
square hectometer
square meter
Symbol
Gy
m/s2
Bq
km2
hm2
m2
Acceptable equivalent
J/kg
1/s
ha (hectare)
The spellings metre and litre are preferred by American Society for Testing and Materials (ASTM); however, er will be
used in the Encyclopedia.
z
Weight is the commonly used term for mass.
Wide use is made of Celsius temperature t defined t T T0 where T is the thermodynamic temperature, expressed in
kelvins, and T0 273:15 K by definition. A temperature interval may be expressed in degrees Celsius as well as in kelvins.
xxxi
xxxii
Quantity
equivalent
*
capacitance
concentration (of amount of substance)
*
conductance
current density
density, mass density
dipole moment (quantity)
*
electric charge, quantity of electricity
electric charge density
electric field strength
electric flux density
*
electric potential, potential difference,
electromotive force
*
electric resistance
*
energy, work, quantity of heat
energy density
*
force
*
frequency
heat capacity, entropy

heat capacity (specific), specific entropy
heat transfer coefficient
*
illuminance
inductance
linear density
luminance
*
luminous flux
magnetic field strength
*
magnetic flux
*
magnetic flux density
molar energy
molar entropy, molar heat capacity
moment of force, torque
momentum
permeability
permittivity
*
power, heat flow rate, radiant flux
*
power density, heat flux density,

irradiance
*
pressure, stress
sound level
specific energy
specific volume
surface tension
thermal conductivity
velocity
viscosity, dynamic
y
Unit
Symbol
Acceptable
farad
mole per cubic meter
siemens
ampere per square meter
kilogram per cubic meter
coulomb meter
coulomb
coulomb per cubic meter
volt per meter
coulomb per square meter
F
mol/m3
S
A/m2
kg/m3
Cm
C
C/m3
V/m
C/m2
C/V
volt
ohm
megajoule
kilojoule
joule
electron volty
kilowatt houry
joule per cubic meter
kilonewton
newton
megahertz
hertz
joule per kelvin
joule per kilogram kelvin
watt per square meter
kelvin
lux
henry
kilogram per meter
candela per square meter
lumen
ampere per meter
weber
tesla
joule per mole
joule per mole kelvin
newton meter
kilogram meter per second
henry per meter
farad per meter
kilowatt
watt
V
V
MJ
kJ
J
eVy
kWhy
J/m3
kN
N
MHz
Hz
J/K
J/(kgK)
W/(m2K)
watt per square meter

megapascal
kilopascal
pascal
decibel
joule per kilogram
cubic meter per kilogram
newton per meter
watt per meter kelvin
meter per second
kilometer per hour
pascal second
millipascal second
W/m2
MPa
kPa
Pa
dB
J/kg
m3/kg
N/m
W/(mK)
m/s
km/h
Pas
mPas
lx
H
kg/m
cd/m2
lm
A/m
Wb
T
J/mol
J/(molK)
Nm
kgm/s
H/m
F/m
kW
W
A/V
g/L; mg/cm3
As
W/A
V/A
Nm
kgm/s2
1/s
lm/m2
Wb/A
cdsr
Vs
Wb/m2
J/s
N/m2
This non-SI unit is recognized as having to be retained because of practical importance or use in specialized fields.
Quantity
viscosity, kinematic
Unit
square meter per second
square millimeter per second
cubic meter
cubic decimeter
cubic centimeter
1 per meter
1 per centimeter
wave number
Symbol
m2/s
mm2/s
m3
dm3
cm3
m1
cm1
xxxiii
Acceptable equivalent
L(liter)
mL
In addition, there are 16 prefixes used to indicate order of magnitude, as follows:
Multiplication factor
1018
1015
1012
109
108
103
102
10
101
102
103
106
109
1012
1015
1018
Prefix
exa
peta
tera
giga
mega
kilo
hecto
deka
deci
centi
milli
micro
nano
pico
femto
atto
Symbol
E
P
T
G
M
k
ha
daa
da
ca
m
m
n
p
f
a
Note
Although hecto, deka, deci, and centi are

SI prefixes, their use should be avoided
except for SI unit-multiples for area and
volume and nontechnical use of
centimeter, as for body and clothing
measurement.
For a complete description of SI and its use the reader is referred to ASTM E 380.
CONVERSION FACTORS TO SI UNITS

A representative list of conversion factors from non-SI to SI units is presented herewith. Factors are given to four significant
figures. Exact relationships are followed by a dagger (y). A more complete list is given in ASTM E 380-76 and ANSI Z210.
1-1976.
To convert from
acre
angstrom
are
astronomical unit
atmosphere
bar
barrel (42 U.S. liquid gallons)
Btu (International Table)
Btu (mean)
Bt (thermochemical)
bushel
calorie (International Table)
calorie (mean)
calorie (thermochemical)
centimeters of water (39.2 8F)
centipoise
centistokes
To
square meter (m2)
meter (m)
square meter (m2)
meter (m)
pascal (Pa)
pascal (Pa)
cubic meter (m3)
joule (J)
joule (J)
joule (J)
cubic meter (m3)
joule (J)
joule (J)
joule (J)
pascal (Pa)
pascal second (Pas)
square millimeter per second (mm2/s)
Multiply by
4:047 103
1:0 1010y
1:0 102y
1:496 1011
1:013 105
1:0 105y
0.1590
1:055 103
1:056 103
1:054 103
3:524 102
4.187
4.190
4.184y
98.07
1:0 103y
1.0y
xxxiv
To convert from
cfm (cubic foot per minute)
cubic inch
cubic foot
cubic yard
curie
debye
degree (angle)
denier (international)
dram (apothecaries)
dram (avoirdupois)
dram (U.S. fluid)
dyne
dyne/cm
electron volt
erg
fathom
fluid ounce (U.S.)
foot
foot-pound force
foot-pound force
foot-pound force per second
footcandle
furlong
gal
gallon (U.S. dry)
gallon (U.S. liquid)
gilbert
gill (U.S.)
grad
grain
gram force per denier
hectare
horsepower (550 ftlbf/s)
horsepower (boiler)
horsepower (electric)
hundredweight (long)
hundredweight (short)
inch
inch of mercury (32 8F)
inch of water (39.2 8F)
kilogram force
kilopond
kilopond-meter
kilopond-meter per second
kilopond-meter per min
kilowatt hour
kip
knot international
lambert
league (British nautical)
league (statute)
light year
liter (for fluids only)
maxwell
micron
mil
mile (U.S. nautical)
mile (statute)
mile per hour
To
cubic meter per second (m3/s)
cubic meter (m3)
cubic meter (m3)
cubic meter (m3)
becquerel (Bq)
coulomb-meter (Cm)
radian (rad)
kilogram per meter (kg/m)
tex
kilogram (kg)
kilogram (kg)
cubic meter (m3)
newton(N)
newton per meter (N/m)
joule (J)
joule (J)
meter (m)
cubic meter (m3)
meter (m)
joule (J)
newton meter (Nm)
watt(W)
lux (lx)
meter (m)
meter per second squared (m/s2)
cubic meter (m3)
cubic meter (m3)
ampere (A)
cubic meter (m3)
radian
kilogram (kg)
newton per tex (N/tex)
square meter (m2)
watt(W)
watt(W)
watt(W)
kilogram (kg)
kilogram (kg)
meter (m)
pascal (Pa)
pascal (Pa)
newton (N)
newton (N)
newton-meter (Nm)
watt (W)
watt(W)
megajoule (MJ)
newton (N)
meter per second (m/s)
candela per square meter (cd/m2)
meter (m)
meter (m)
meter (m)
cubic meter (m3)
weber (Wb)
meter (m)
meter (m)
meter (m)
meter (m)
meter per second (m/s)
Multiply by
4:72 104
1:639 104
2:832 102
0.7646
3:70 1010y
3:336 1030
1:745 102
1:111 107
0.1111
3:888 103
1:772 103
3:697 106
1:0 106y
1:00 103y
1:602 1019
1:0 107y
1.829
2:957 105
0.3048y
1.356
1.356
1.356
10.76
2:012 102
1:0 102y
4:405 103
3:785 103
0.7958
1:183 104
1:571 102
6:480 105
8:826 102
1:0 104y
7:457 102
9:810 103
7:46 102y
50.80
45.36
2:54 102y
3:386 103
2:491 102
9.807
9.807
9.807
9.807
0.1635
3.6y
4:448 102
0.5144
3:183 103
5:559 102
4:828 103
9:461 1015
1:0 103y
1:0 108y
1:0 106y
2:54 105y
1:852 103y
1:609 103
0.4470
To convert from
To
Multiply by
millibar
millimeter of mercury (0 8C)
millimeter of water (39.2 8F)
minute (angular)
myriagram
myriameter
oersted
ounce (avoirdupois)
ounce (troy)
ounce (U.S. fluid)
ounce-force
peck (U.S.)
pennyweight
pint (U.S. dry)
pint (U.S. liquid)
poise (absolute viscosity)
pound (avoirdupois)
pound (troy)
poundal
pound-force
pound per square inch (psi)
quart (U.S. dry)
quart (U.S. liquid)
quintal
rad
rod
roentgen
second (angle)
section
slug
spherical candle power
square inch
square foot
square mile
square yard
store
stokes (kinematic viscosity)
tex
ton (long, 2240 pounds)
ton (metric)
ton (short, 2000 pounds)
torr
unit pole
yard
pascal (Pa)
pascal (Pa)
pascal (Pa)
radian
kilogram (kg)
kilometer (km)
ampere per meter (A/m)
kilogram (kg)
kilogram (kg)
cubic meter (m3)
newton (N)
cubic meter (m3)
kilogram (kg)
cubic meter (m3)
cubic meter (m3)
pascal second (Pas)
kilogram (kg)
kilogram (kg)
newton (N)
newton (N)
pascal (Pa)
cubic meter (m3)
cubic meter (m3)
kilogram (kg)
gray (Gy)
meter (m)
coulomb per kilogram (C/kg)
radian (rad)
square meter (m2)
kilogram (kg)
lumen (lm)
square meter (m2)
square meter (m2)
square meter (m2)
square meter (m2)
cubic meter (m3)
square meter per second (m2/s)
kilogram per meter (kg/m)
kilogram (kg)
kilogram (kg)
kilogram (kg)
pascal (Pa)
weber (Wb)
meter (m)
1:0 102
1:333 102y
9.807
2:909 104
10
10
79.58
2:835 102
3:110 102
2:957 105
0.2780
8:810 103
1:555 103
5:506 104
4:732 104
0.10y
0.4536
0.3732
0.1383
4.448
6:895 103
1:101 103
9:464 104
1:0 102y
1:0 102y
5.029
2:58 104
4:848 106
2:590 106
14.59
12.57
6:452 104
9:290 102
2:590 106
0.8361
1:0y
1:0 104y
1:0 106y
1:016 103
1:0 103y
9:072 102
1:333 102
1:257 107
0.9144y
xxxv
H
continued
HYDROCEPHALUS, TOOLS FOR DIAGNOSIS

AND TREATMENT OF
CSF Circulation. Unlike other organ systems, the brain

and the spinal cord are unique in being bathed in a clear
fluid called cerebrospinal fluid. The exact role that it plays
in maintaining the necessary environment for the functioning of the nervous system is unclear. It has been
ascribed a role in providing nutrition, removing excess
waste, circulating neurotransmitters, maintaining the
necessary electrolyte environment, and acting as a shock
absorber against trauma.
The distribution of nutrients, or neurotransmitters, and
removal of waste products of metabolism, is an unlikely
function of CSF, because these chemicals are present in
very low concentrations in the CSF. The main function of
CSF is to provide buoyancy to support the brain and act as
a cushion against trauma. The normal brain weighs about
1500 g; however, supported by the buoyancy of the CSF, its
apparent weight is reduced to about 50 g in the cranium.
Support for its role in cushioning the brain and spinal cord
against trauma comes from clinical conditions like severe
spinal canal stenosis. The CSF cushion around at the site of
stenosis is markedly reduced. As a result, spinal cord injury
often occurs even with minor trauma as the shock waves are
directly transmitted from the bone to the spinal cord.
Cerebrospinal fluid is made through a complex process
that occurs in the cells of the choroid plexus, which
lines the margin of the four fluid-filled spaces in the brain
called the ventricles. First, an ultrafilterate of plasma is
formed in the connective tissue surrounding the choroidal
capillaries. Next, this is converted into a secretion by
carbonic anhydrase enzyme present in the choroids epithelium. The CSF is made at a fairly constant rate of about 10
mL/h. Most of the CSF is made in the choroids plexus of the
lateral ventricles. Roughly, 20% of the CSF comes from the
ventricular walls. As most CSF is made in the lateral
ventricles, it is traditionally believed that the CSF bulk
flow occurs from the lateral ventricles to the third ventricle,
fourth ventricle, and then through the foramen of Magendie and Lushka into the cerebello-pontine cistern and on to
the surface of the brain and spinal cord (Fig. 1). A fifth of
the CSF runs down around the spinal cord and then back to
the cranial subarachnoid space.
The CSF is absorbed by the cells of the arachnoid
granulations (13). These are present in the superior sagittal sinus. The process involves pinocytosis of a small
quanta of CSF, on the subarachnoid side of the granulations, and discharge into the blood on the venous side. The
process is driven by a pressure difference of at least 5 mm
Hg between the subarachnoid CSF and the superior sagittal sinus. A small proportion of CSF is also absorbed along
the perivascular spaces and along the nerve sheaths exiting the spinal canal (14).
This traditional view has been recently challenged.
Johnston et al. in experimental and cadaveric studies have
demonstrated that a large amount of CSF is present
around the olfactory nerve and the cribriform plate area
SANDEEP SOOD
ANDERS EKLUND
NOAM ALPERIN
University of Illinois at Chicago
Chicago, Illinois
INTRODUCTION
Epidemiology
A congenital form of hydrocephalus occurs in roughly 50
in 100,000 live births (6). Hydrocephalus may also be
acquired later in life as a result of a brain tumor, following
meningitis, trauma, or intracranial hemorrhage. It has been
estimated that prevalence of shunted hydrocephalus is
about 40/100,000 population in the United States (7).
Untreated hydrocephalus has a poor natural history with
a mortality rate of 2025% and results in severe physical
and mental disabilities in survivors (8,9). There has been a
significant reduction in mortality and morbidity with use of
shunting. However, shunting is associated with a high
failure rate; a 40% failure rate occurs within the first year
after shunting (10). Advances in the technology have lead to
the development of a diverse type of shunt systems to
circumvent problems related to long-term shunting, such
as obstruction, infection, and overdrainage. Yet, studies done
to evaluate these devices have not shown a significant longor short-term benefit from their use compared with the
conventional devices (10). It is estimated that, during the
year 2000, the cost associated with shunting exceeded one
billion dollars in the United States alone (11). Shunt replacement accounted for 43% of shunt procedures. Endoscopic
surgery has provided an alternative strategy in patients with
obstructive hydrocephalus. However, limited data in the
literature suggest that long-term survival of third ventriculostomy is not significantly superior to that of a shunt (12).
Physiology
The CSF flow in the craniospinal system is influenced
by two separate processes: (1) the circulation of the CSF
from its formation sites to its absorption sites (i.e., bulk
flow) and (2) an oscillatory (back and forth) flow during
the cardiac cycle (pulsatile flow). The first process governs
the overall volume of CSF and thereby influences intracranial pressure (ICP). The second process, the oscillatory
movement of the CSF within the craniospinal compartments, is caused by the pulsatile blood flow entering
and leaving the intracranial compartment during the
cardiac cycle. These two processes occur over different
time scales; circulation and replenishing of CSF occurs
over minutes, whereas the time scale of the pulsatile CSF
flow is milliseconds.
1
HYDROCEPHALUS, TOOLS FOR DIAGNOSIS AND TREATMENT OF
can occur from obstruction to the egress of CSF from the

ventricles. This is referred to as obstructive or noncommunicating hydrocephalus. It may also result from impairment
of absorption of the CSF at the level of the arachnoid villi or
increased resistance to the flow of CSF in the subarachnoid
spaces from fibrosis and scarring related to meningitis or
previous subarachnoid hemorrhage. This is referred to as
communicating hydrocephalus. Irrespective of the cause,
the accumulation of CSF has two consequences. It results in
an increase in the pressure in the cranium and may cause
diltation of the ventricles (ventriculomegaly).
Figure 1. CSF is mainly formed by the choroids plexus in the

lateral, third (1), and fourth (4) ventricles. The fluid flows (in the
direction of the arrows) from lateral ventricles through the foramen
of Monro (6) into the third ventricle. CSF then passes through
the aqueduct of Sylvius (2) into the fourth ventricle (3) and exits the
fourth ventricle through the foramen of Luschka and Magendie (5)
into the cisterna magna and the basal cisterns. The flow is then
into the subarachnoid space over the surface of the brain and about
the spinal cord. Finally, the fluid is absorbed through the tufts of
arachnoid (arachnoid villi) into the sagittal sinus.
and drains into the lymphatic system of the face (15,16).

Others believe that CSF may be absorbed directly at the
level of the capillaries and perivascular spaces (17).
CSF Pulsations. The pulsatile back and forth movement
of CSF between the cranium and the spinal canal with each
heartbeat plays a role in modulating the pulsatile cerebral
blood flow. Blood flow in the arteries leading blood to the
brain is pulsatile, whereas the pulsatility of blood flow in
the cerebral veins is considerably attenuated. The displacement of CSF into the spinal canal during the systolic
phase helps accommodate the temporary increase in blood
volume in the intracranial space, which otherwise has only
a limited capacity to accommodate additional volume due
to the limited compliance of the intracranial compartment.
Reducing the pulsatility of the blood flow through the brain
may play a role in diffusion of nutrients to the brain cells
from the blood and of waste products from the brain cell to
the blood through a less pulsatile flow at the level of the
capillaries. As discussed later, MRI measurements of
the pulsatile arterial, venous, and CSF flows, to and from
the cranium, can now be used to measure intracranial
compliance and pressure (ICP), noninvasively. As one of
the main roles of shunting is to protect the brain from
increased ICP, diagnostic noninvasive measurement of
ICP may aid in management decisions in hydrocephalous.
Pathophysiology
Hydrocephalus occurs if there is a mismatch between the
CSF production and the absorption. Accumulation of CSF
Increase in Intracranial Pressure: Maintaining

normal intracranial pressure is important for the
functioning of the brain. The pressure in the intracranial cavity increases exponentially with an
increase in the total volume of its content (brain
tissue, blood, and the CSF) (18). Therefore, increase
in intracranial volume, due to uncompensated accumulation of CSF, increases ICP and reduces intracranial compliance. Compliance quantifies the ability of a
compartment to accommodate increase in volume for
a given increase in pressure and is defined as the ratio
of the changes in volume and pressure:
Compliance
Dv
Dp
where, Dv is change in volume and, Dp is the change in

pressure. Intracranial compliance decreases with
increased ICP because of the exponential relationship
between ICP and intracranial volume (ICV).
Normal ICP is about 15 mm Hg in an infant and up to 20
mm Hg in an adult. It is measured by inserting a needle into
the spinal canal and recording the pressure using a manometer or by placing a catheter with miniature strain gauge
transducer at its distal tip (Codman, Raynham, MA;
Camino, Integra LifeSciences, Plainsboro, NJ) directly into
the brain parenchyma or the ventricles through a small
twist drill hole in the skull. Noninvasive means for measurement of ICP would be important for diagnosis and management of hydrocephalus. Over the last several decades,
different approaches have been attempted (19). A method
based on measurements of CSF and blood flows to and from
the brain by MRI is described in more detail in this article.
Increase in ICP can affect the brain in two ways. First, it
reduces perfusion of blood into the brain due to the reduced
cerebral perfusion pressure (i.e., arterial pressure minus
ICP). Depending on the severity and duration, it may
result in chronic ischemia causing impairment in higher
mental functions, developmental delay in children, or an
acute ischemic injury and stroke. Second, rise in pressure
in any one of the compartments in the cranium, formed by
the tough dural falx in the midline and the tentorium
between the cerebral hemispheres superiorly and the cerebellum inferiorly, forces the brain to herniate. This often
leads to infarction of the brain stem and death.
Symptoms: Clinically, patients who have elevated
ICP generally present with typical symptoms. Headache is the most common. It occurs especially in the
early hours of the morning in initial stages. Low

respiratory rate during sleep results in buildup of blood
CO2 and vasodiltation. This aggravates the increased
ICP in the early stages of the disease. Vomiting is the
next common symptom and probably results either
from the distortion of the brain stem vomiting center
or its ischemia. Vomiting is often associated with
retching and rapid respiration that lowers the blood
CO2 level. This in turn leads to vasoconstriction and
lowers the ICP and often results in a transient relief in
headaches. Diplopia or double vision is also commonly
encountered in a setting of increased ICP. It is a result
of stretch of the sixth cranial nerve, which controls the
abduction of the eyes. Weakness of ocular abduction
disturbs the normal axial alignment of the two eyes
resulting in defective fusion of the two images by the
brain. Blurring of vision and visual loss may occur in
patients with long-standing intracranial hypertension.
This results from edema of the optic nerve head as the
axoplasmic flow in the neurons of the optic nerve is
impaired by the high ICP that is transmitted to the
nerve through the patent nerve sheath. Hearing
deficits related to similar effect on the cochlea are,
however, less frequent. Lethargy or sleepiness is frequently observed in patients with high ICP and is
probably from a combination of decreased cerebral
perfusion and distortion of the brain stem.
Ventricular Enlargement: Depending on the
pathophysiology of hydrocephalus, CSF may accumulate only in the ventricles as in obstructive hydrocephalus or in both the ventricles and the subarachnoid
space in communicating hydrocephalus. The increased
pressure within the ventricle is transmitted to the
periventricular region and results, over time, in loss
of neurons, increase in periventricular interstitial fluid,
and subsequent gliosis with loss of white matter (20).
When onset of hydrocephalus occurs early in infancy,
before the skull sutures have closed, the enlarging ventricles are associated with a progressive increase in head
circumference and developmental delay. In later childhood
and adults, the increasing ventricular size is associated
with symptoms of increased ICP. However, ventricular
enlargement may also occur with normal mean ICP, in
the elderly patients (21). This is referred to as normal
pressure hydrocephalus (NPH). The enlarging ventricle
stretches the periventricular nerve fibers. The patient
presents not with signs of increase in ICP but with progressive gait ataxia, bladder incontinence, and dementia.
Similar presentation may also be observed in adolescents
with aqueductal stenosis and obstructive hydrocephalus.
These patients with compensated long-standing hydrocephalus have been referred to as long-standing hydrocephalus of adults (LOVA) (22).
It is not clear why the ventricles enlarge preferentially,
compared with the subarachnoid space, even though the
pressure distributes equally in a closed system. It has been
argued that, rather than the actual mean ICP, it is the pulse
pressure that determines ventricular diltation. Di Rocco et
al. (23) have shown that ventricular enlargement could be
induced by an intraventricular pulsatile balloon with a high
pulse pressure, despite the mean ICP being normal. It may

be argued that in a pulsatile system, it is the root mean
square (RMS) of the pressure, rather than the mean pressure, that is the cause of enlarged ventricles. It has been
suggested that, in communicating hydrocephalus, decrease
in compliance may be responsible for preferential transmission of the pulsations to the ventricles (24,25). However,
others have shown that in acute or chronic communicating
hydrocephalus, the pulse pressure and the pressure waveforms in the SAS and the ventricles are similar (26,27).
Alternative explanations offered are that the pia over the
cortical surface is more resilient than ependyma that lines
the ventricular wall; the venous pressure in the periventricular region is lower, making it more deformable than the
subcortical area (28).
DIAGNOSTIC METHODS
Measurement of Resistance to CSF Reabsorption
The hydrodynamics of the craniospinal system is governed
by patient-specific properties like CSF formation
rate, CSF reabsorption resistance (historically termed
as outflow resistance), venous pressure in the sinus, and
craniospinal compliance. Together with the periodic variations in ICP, due to blood volume variation from the
heartbeat and vasomotion, these properties describe the
CSF dynamics, which provide the working environment of
the brain. When this environment is disturbed, it affects
the function of the brain resulting in the clinical symptoms
of hydrocephalus. After shunting, symptoms are
often eliminated or reduced. It shows that a clinical
improvement can be accomplished by actively changing
the brains working environment. This link among CSF
dynamics, brain function, symptoms, and shunting has
made researchers look for CSF dynamical means to identify patients that would benefit from a shunt surgery.
Outflow resistance has been suggested as a strong predictive
parameter in communicating hydrocephalus. Invasive infusion tests in conjunction with a mathematical model of the
craniospinal system can be used to estimate CSF absorption
rate. The most accepted model for the system hydrodynamics
has been proposed by Marmarou (29).
The basic assumptions for the model are as follows:
CSF reabsorption rate is linearly dependent on the
difference between the intracranial and venous pressures (the outflow resistance describes this linear
relationship)
A pressure-dependent compliance
A constant formation rate of CSF, independent of ICP
The model can be displayed as an electrical analogy
(Fig. 2). The model is described mathematically as a differential equation of the time-dependent ICP as a function
of external infusion and the governing physical parameters:

dPIC t
K
K Pr
2
PIC t 0
P t K Iinfusion t
Rout
dt
Rout IC
2
Iabsorption
Itotal
Istorage
Rout
C=1/(K PIC )
Iinfusion
fusion
Iformation
PIC
(ICP)
Psinus
Figure 2. Model of the dynamics of the CSF system. Iformation is

the CSF formation rate, C is the pressure-dependent compliance
described by the elastance parameter K, Rout is outflow resistance,
Psinus is the venous pressure in the sinus, and PIC is the
intracranial pressure. Iinfusion is the option of external infusion
of artificial CSF.
where Rout is outflow resistance, PIC is the intracranial

pressure, Pr is the ICP at rest, and K is the elastance.
Estimation of CSF outflow resistance and the other
system parameters requires perturbation of the system
steady state by infusion of fluid into the craniospinal
system, either through a lumbar or a ventricular route.
Typically, one or two needles are placed in the lumbar
canal. When two needles are used, one is connected to a
pressure transducer for continuous recording of the
dynamic changes in ICP following the infusion, and the
other one for infusion and/or withdrawal of the fluid.
Different protocols of infusion will lead to unique mathematical solutions. In addition to the resting pressure (also
refereed to as opening pressure), which always is determined during these investigations, it is generally believed
that the outflow resistance is the clinically most important
parameter, but compliance has also been proposed as a
predictor for outcome after shunting.
Bolus Infusion. An example of ICP recoding during the
bolus infusion test is shown in Fig. 3. The approach is to
first determine the compliance from the ratio of the injected
volume and the magnitude of pressure increase (30). A
pressure volume index (PVI), which describes compliance,
is calculated through the expression:
PVI
DV
logPp =P0
where DV is the infused volume, Pp is the peak pressure

and P0 is the initial pressure just before the infusion. The
next step is to determine Rout from the spontaneous relaxation curve when the ICP returns toward the resting pressure (Fig. 3). Solving the differential equation for the
relaxation phase after the bolus infusion gives the following expression for Rout as a function of time (31):
Rout
tP0

Pt =P p P p P0
PVI log
Pt P0
where t is the time in seconds after the bolus and Pt is the

measured pressure at time t on the relaxation curve. From
each bolus, a number of values of Rout are calculated and
averaged, for example at t 1 min, 1.5 min, and 2 min. The
bolus procedure is usually repeated a couple of times for
increased measurement reliability.
Constant Pressure Infusion. In this infusion protocol,
several constant ICP levels are created. This is done by
using a measurement system that continuously records the
ICP and regulates it by controlling the pump speed of an
infusion pump (Fig. 4) (32). The net infusion rate needed to
sustain ICP at each pressure level is determined, and a flow
versus pressure curve is generated (Fig. 4). Using linear
regression, the outflow resistance is then determined from
the slope of that curve (33), because at steady state, the
differential equation reduces to
Iinf
1
Pr
P
Rout IC Rout
where PIC is the mean intracranial pressure on each level,

Pr is the resting pressure, and Iinf is the net infusion flow at
each level.
The constant pressure method can also be used to
estimate the CSF formation rate. This is done by lowering
the ICP beneath the venous pressure, i.e., below 5 mm Hg.
At that ICP, no CSF reabsorption should take place. Therefore, the net withdrawal of CSF needed to sustain that
constant pressure level should equal the formation rate.
3
Constant Flow Infusion. In this method, both the static
and the dynamic behavior of the CSF system can be used to
estimate outflow resistance (34). In a steady-state analysis
Rout can be calculated from the stable ICP value associated
with a certain constant infusion rate. Rout is then estimated
by the following expression:
Rout;stat
Figure 3. ICP curve from a bolus injection of 4 mL. Figure from

Marmarou et al. (30).
Plevel Pr
Iinf
where Rout,stat is a static estimation of Rout, Plevel is the new

equilibrium pressure obtained at the constant infusion
rate, Pr is the resting pressure, and Iinf is the infusion rate
(Fig. 5).
Rout can also be estimated from the dynamic phase
during the pressure increases toward the new equilibrium
(Fig. 5). This procedure will also give an estimate of the
craniospinal compliance (elastance). The differential equation is now solved for a condition of a constant infusion rate,
and the solution is fitted against the recorded pressure

160
45
140
ICP
Net infusion
Flow = 0.13 ICP - 0.93
120
30
100
25
80
20
60
Flow [ml/min]
ICP [mm Hg]
35
Vol [ml]
40
15
40
10
20
5
0
00:00
00:20
00:40
01:00
01:20
0
01:40
Time [h:min]
0
15
20
25
ICP [mm Hg]
30
35
Figure 4. Pressure and flow curves for the constant pressure method. Left graph shows typical data
for ICP and infused CSF volume versus time. Right graph shows the mean pressure and flow points
determined from each steady-state level. Outflow resistance corresponds to the inverse of the slope.
curve. The time-dependent pressure increase is described

by the following expression:

r P0
Pr P0
Iinf RPout;dyn

# P0
Pt
7
"
Pr P0
Rout;dyn
K
Iinf e
Pr P0
Rout;dyn Iinf
where K is the elastance and P0 is a reference pressure that

is suggested to be equal to venous sinus pressure. Fitting
against data will result in estimations of the unknown
parameters Rout,dyn, K, and P0.
In summary, CSF infusion tests are conducted to reveal
parameters describing the hydrodynamics of the craniospinal system. An active infusion of artificial CSF is performed, and the resulting ICP response is recorded, and
parameters such as outflow resistance, compliance, formation rate, and the venous sinus pressure are then estimated
based on a proposed mathematical model. Outflow resistance values determined with the bolus method are usually
lower than the values determined with the constant infusion and constant pressure methods. The reason for this
difference is not well understood at this time. Determina-
tion of Rout is often used as a predictive test in hydrocephalus, and it has been stated that if the outflow resistance
exceeded a certain threshold, it is an excellent predictor of
clinical improvement after shunting (35). In a recent guideline for idiopathic normal pressure hydrocephalus, measurement of Rout is included as a supplementary test for
selecting patients suitable for shunt surgery (36).
DIAGNOSIS WITH IMAGING
Cross-sectional imaging is routinely used in the diagnosis
of hydrocephalous. CSF spaces are well visualized with CT
and MRI. In CT images, CSF spaces appear darker due to
the lower atomic density of the CSF compared with that of
brain tissue. MRI provides an excellent soft-tissue contrast
resolution and is considered the primary imaging modality
for brain imaging. With MRI, CSF spaces can appear either
darker or brighter compared with its surrounding tissues
depending on the imaging technique. An example of a CT
image and MRI images demonstrating abnormally large
CSF spaces is shown in Fig. 6. Cross-sectional imaging
enables quantitative assessment of the CSF spaces as well
Figure 5. ICP data from a constant

infusion investigation. Figure modified
from Czosnyka et al. (34).
Figure 6. An example of a (a) CT image and (b and c) MRI images demonstrating abnormally large
CSF spaces. The appearance of CSF in MRI depends on the technique used to acquire the images; its
(b) dark with a T1 technique and (c) bright with a T2 technique.
as 3D reconstruction of the geometry of the ventricular

system. The 3D model is obtained by segmentation of the
CSF spaces in each of the 2D slices. An example of 3D
models of the ventricular system from MRI data demonstrating normal size ventricles and enlarged ventricles are
shown in Fig. 7a and b, respectively.
MRI-based motion-sensitive techniques capable of imaging flow are gaining an important role in the diagnosis
of hydrocephalus. In particular, dynamic phase-contrast
techniques provide images of velocities (velocity-encoded
images). The degree of brightness in these images is proportional to the direction and the speed of the moving fluid
or tissue. Dynamic (cine) phase contrast images are used to
visualize the back and forth flow through the different CSF
pathways. The cine phase contrast MRI (PCMRI) technique is also used to derive quantitative parameters such as
CSF volumetric flow rate through the aqueduct of Sylvius,
from which the CSF production rate in the lateral ventricles can be estimated (37), and intracranial compliance and
pressure (19,30).
MRI-Based Measurement of Intracranial Compliance and
Pressure
The noninvasive measurement of compliance and pressure
uses the cardiac pulsations of the intracranial volume and
Figure 7. Volume rendering of the

CSF spaces inside the brain (i.e., ventricles) generated using segmented
MRI data from a (left) healthy volunteer and from a (right) hydrocephalic patient.
pressure (30,38). This method is the noninvasive analogs to

the measurement of intracranial compliance with the previously described bolus infusion method where the volume
and pressure changes are calculated from the MRI measurements of CSF and blood flows to and from the brain.
Intracranial elastance, i.e., a change in pressure due to a
small change in volume, or the inverse of compliance, is
derived from the ratio of the magnitudes of the changes in
volume and pressure, and the pressure is then derived
through the linear relationship between elastance and
pressure. The MRI method measures the arterial, venous,
and CSF flows into and out of the cranial vault. A smallvolume change, on the order of 1 mL, is calculated from the
momentary differences between inflow and outflow at each
time points in the cardiac cycle. The pressure change is
proportional to the pressure gradient change, which is
calculated from time and spatial derivatives of the CSF
velocities using fluid dynamics principles.
A motion-sensitive MRI technique, cine phase contrast, provides a series of images where the value at each
picture element is proportional to the velocity at that
location. The phase contrast MRI technique is based on
the principle that the precession frequency of the protons
is proportional to the magnetic field strength. Therefore,
velocity can be phased-encoded by varying the magnetic
field in space and time, i.e., generating magnetic field
Figure 8. (a) Anatomical midsagittal T1-weighted MR image

showing the location of the axial plane
used for CSF flow measurement (dark
line). (b and c) Phase-contrast MRI
images of CSF flow in the spinal canal.
(b) CSF flow during systole. (c) CSF flow
during diastole. The pixel values in these
images are proportional to velocities in a
direction perpendicular to the image
plane. Gray-static tissue, whiteoutward flow (caudal direction), and
black-inward flow (cranial direction).
(d) A 3D plot of the CSF velocities
during systole.
gradients. When a gradient field is applied along an axis

for a short time, the protons phase will change based on
its location along that axis. When a bipolar (positive and
then negative) gradient field is applied, the phase of the
stationary protons will increase during the positive portion (lobe) of the bipolar gradient and then will decrease
during the negative lobe. If the lobes were of equal area,
no net phase change would occur. However, moving protons, such as those in the blood or CSF, will experience
different field strength during each lobe due to their
change in position; this will result in a net phase change

proportional to the proton velocity.
Examples of MRI phase contrast images of CSF and
blood flow are shown in Figs. 8 and 9, respectively. The
oscillatory CSF flow between the cranial and the spinal
compartments is visualized in images taken in a transverse
anatomical orientation through the upper cervical spinal
canal. The location of this plane is indicated on a midsagittal scout MR image shown in Fig. 8a. Fig. 8b depicts
outflow (white pixels) during systole, and Fig. 8c depicts
Figure 9. (a) A blood vessel MRI scout image

showing the location of the axial plane for blood
flow measurement (dash line). (b) A phase contrast MRI image of blood flow through that
location. Black pixels indicate arterial inflow,
and white are venous outflow. (c) A 3D plot of
the blood flow velocities.

1400
csf
a
1200
Flow (mL/min)
v
1000
800
600
400
200
0
200
0
200
400
600
800
1000
Time (ms)
Figure 10. The volumetric flows into and out of the cranial vault
during the cardiac cycle derived from the MRI phase contrast
scans. Total arterial inflow (filled circles), venous outflow (open),
and the cranial-to-spinal CSF volumetric flow rate (diamonds)
during the cardiac cycle. Note that arterial inflow is greater
than venous outflow during systole.
inflow (black pixels) during diastole. Fig 9d depicts a 3D

plot of the velocities in a region of interest containing the
CSF space and an epidural vein. The CSF flow is imaged
with a low-velocity encoding, and the faster blood flow
through the neck arteries and veins is imaged using
high-velocity encoding. The location of the imaging plane
used for blood flow measurement is shown in Fig. 9a, and
a velocity encoded image of blood flow is shown in Fig. 9b.
Fig 9c depicts a 3D plot of the velocities in a region of
interest containing the internal carotid and vertebral
arteries and the jugular vein.
Volumetric flow rates are obtained by integration of the
velocities throughout a lumen cross-sectional area. The
total volumetric arterial flow ratethat is, total cerebral
blood flowis calculated directly from the sum of the
volumetric flow through the four vessels carrying blood
to the brain (internal carotid and vertebral arteries). The
venous blood outflow is obtained by summation of the flow
through the jugular veins, and through secondary venous
outflow channels such as the epidural, vertebral, and deep
cervical veins when venous drainage occurs through these
veins. An example of the volumetric flow waveforms for
CSF, arterial inflow, and venous outflow measured in a
healthy volunteer is shown in Fig. 10.
The rate of the time-varying intracranial volume change
(net transcranial volumetric flow rate) is obtained by sub-
tracting outflow rates from inflow rates at each time point.

The intracranial volume change (delta of volume from a
given reference point) is obtained by integrating that waveform with respect to time. Waveforms of the net transcranial volumetric flow rate and the change in the intracranial
volume are shown in Fig. 11.
The magnitude of the change in intracranial pressure
during the cardiac cycle (pulse pressure) is proportional to
that of the CSF pressure gradient waveform. A method to
measure pressure gradient of pulsatile flow in tubes with
MRI was reported by Urchuk and Plewes (39). Pulsatile
pressure gradients are derived from the MRI velocityencoded phase contrast images using the NavierStokes
relationship between pressure gradient and temporal and
spatial derivatives of the fluid velocity for incompressible
fluid in a rigid tube (40). Pressure traces from invasive
recordings obtained invasively in patients with low and
elevated ICP with an intraventricular pressure transducer
and the corresponding CSF pressure gradient waveforms
derived from the MRI measurements of the CSF velocities
at low- and high-pressure states are shown in Fig 12. The
ratio of the magnitude of the pressure and volume changes,
i.e., intracranial elastance, is then expressed in terms of
MR-ICP based on the linear relationship between elastance and ICP.
DEVICES FOR TREATMENT

Despite significant advances in understanding of the
pathophysiology of hydrocephalus, the gold standard for
the treatment of hydrocephalus still continues to be CSF
diversion through a tube shunt to another body cavity.
Unfortunately, treatment with CSF shunts is associated
with multiple complications and morbidity. The rate of
shunt malfunction in the first year of shunt placement is
40%, and, thereafter, about 10% per year. The cumulative
risk of infection approaches 20% per person although the
risk of infection per procedure is only 58% (41). The
technological advances in shunt valve designs and materials have had only a marginal impact on the rate of complications. Third ventriculostomy has become popular in
recent years for management of obstructive hydrocephalus, but many questions about its long-term permanence
remain controversial. Choroid plexectomy (42,43) aimed at
arresting hydrocephalus by reducing CSF production or
pharmacotherapy with similar intentions have had very
limited success in selected patients.
0.8
250
ICVC (mL)
Figure 11. (Left) The MRI-derived

net transcranial volumetric flow rate
waveform. (Right) The intra cranial
volume change during the cardiac
cycle derived by integrating the net
transcranial volumetric flow waveform
on the left. Note that the maximal
volume change in this subject is 0.5 mL.
Flow (mL/min)
500
0
250
500
dV = 0.5mL
0.4
0
0.4
200
400
600
Time (ms)
800
1000
200
400 600
Time (ms)
800
1000
Figure 12. Invasive pressure traces (top) obtained with an intra-ventricular catheter from two
patients with (left) low and (right) elevated ICP. The corresponding MRI-derived CSF pressure
gradients are shown at the bottom.
Nonobstructive Hydrocephalus
No treatment other than CSF diversion has been effective in
management of this form of hydrocephalus. The CSF may be
diverted from the ventricles through a catheter that runs in
the subcutaneous tissue into the abdominal cavity where it
is absorbed by the peritoneum (ventriculo- peritoeal shunt)
(Fig. 13). It may also be diverted from the spinal subarachnoid space by a lumbar shunt that diverts it to the peritoneum (Lumbar-peritoneal shunt). Lumbar CSF diversion
avoids the potential risk of brain injury by the ventricular
catheter. Lumbar shunts have a lower risk of obstruction
and infection (44) but are more prone to malfunction from
mechanical failures (45), and, the development of hind brain
herniation, over a period of time, has been well documented
(46,47). Evaluation of a lumbar shunt for function is more
cumbersome than that of a ventricular shunt. The lumbar
shunt is usable in patients with communicating hydrocephalus, small ventricles, and patients who have had multiple ventricular shunt malfunctions.
In patients who cannot absorb CSF from the peritoneum
due to scarring from previous operations or infections, the
CSF may be diverted to the venous system through a
catheter placed at the junction of superior vena cava and

the right atrium (ventriculo/lumbar-atrial shunt).
A typical shunt system consists of three parts (Fig. 14).
First, the proximal catheter, i.e, the catheter, is inserted into
the ventricle or the lumbar subarachnoid space. Second, the
valve controls the amount of CSF that flows through the
shunt system, and third, the distal catheter drains the CSF
from the valve to the peritoneum or the atrium.
Proximal Catheter
Three basic types of proximal catheter designs are
available: simple with multiple perforations (Codman,
Raynham, MA; PS Medical, Goleta, CA), simple Flanged
(Heyer-Schulte), Integra, Plainsboro, NJ; Anti-Blok
(Phoenix Vygon Neuro, Valley Forge, PA) with receded
perforations. The last two have been designed to minimize
the growth of choroid plexus into the perforations and
causing obstruction. There is no controlled study to suggest that these two designs are in any way superior to
simple perforations. The flanged catheters can get stuck,
as choroid plexus grows around it, making removal of an
obstructed catheter difficult (48).
10
choroid plexus is torn while forcefully removing the catheter. Delayed subarachnoid hemorrhage from rupture of
pseudoaneurysm resulting from diathermy of a catheter
close to anterior cerebral artery has been reported (59). At
times, if the ventricular catheter is severely stuck, it is
advisable to leave it in position but occlude it by a ligature
and clip. This may become necessary as sometimes an
occluded catheter may become unstuck over time and begin
to partially function, resulting in formation of subgaleal
CSF collection. Replacing a new catheter into the ventricle
in patients with small or collapsed ventricles can be sometimes challenging. In most instances, after removal of the
old catheter, the new catheter can be gently passed into the
ventricle through the same tract. Frameless stereotaxis
(StealthStation, Medtronics, Goleta, PA) is now available
and may offer an alternative to cumbersome and timeconsuming frame-based stereotactic catheter placement
(52).
Valve
Figure 13. The shunt consists of three parts. The ventricular

catheter enters the skull through a burr hole in the skull and
passes through the brain into the lateral ventricle. It is connected
to the valve that is placed in the subcutaneous tissue of the scalp.
The valve in turn is connected to the distal catheter that runs in
the subcutaneous tissue to enter the peritoneal cavity of the
abdomen as shown in the inset (ventriculo-peritoneal shunt) or
into the jugular vein and through it to the superior vena cava
(ventriculo-atrial shunt).
Placement of the proximal catheter has generated considerable controversy in the literature (4851). More
recently, endoscopic placement of the proximal catheter
into the frontal horn, away from the choroid plexus, has
been advocated to minimize proximal malfunction
(3,52,53). Again no controlled study has been done to
confirm whether placement of the proximal catheter into
frontal or occipital horn is superior to placement in the
body of the lateral ventricle. Often catheters that are
grossly malpositioned may continue to work, whereas
those that are well positioned may fail. The choice of the
site, frontal or parietal, may be made on the basis of the
above although some studies have suggested a higher
incidence of seizure with catheters placed via a frontal
burr-hole (49). A study to evaluate use of endoscope to place
the shunt catheter in the frontal horn failed to show any
benefit (54). This suggests that no matter where the catheter is placed, the flow of CSF toward the catheter causes the
choroids plexus to creep toward the catheter, ultimately
causing ingrowth and obstruction of the catheter (55).
To remove an obstructed catheter, intraluminal coagulation of the choroid plexus is done using a stylet and lowvoltage diathermy, at the time of shunt revision (5658).
Massive intraventricular hemorrhage may occur if the
The valve regulates the amount of CSF that is drained. The

aim is to maintain normal ICP. The simplest valves are
differential pressure valves. The CSF drainage in these
valves is based on the pressure difference between the
proximal and the distal ends. Three major configurations
are available (Fig. 14): diaphragm, slit valve, and ballspring mechanism in different pressure ranges (low, medium, and high). Recently, valves in which the pressure
setting can be changed with a magnetic wand have become
available. These programmable valves allow pressure
changes over different pressure ranges based on the manufacturer. The pressure setting on the valve can be ascertained by X ray of the head in the Medos valve (Codman,
Raynham, MA) or using a magnetic wand in the Strata
valve ( Medtronics, Goleta, CA). To prevent inadvertent
changes in the valve setting by stray magnetic fields, the
Polaris valve (Sophysa, Costa Mesa, CA) has an ingenious
locking mechanism that allows changes only if the magnetic field has a certain configuration.
Slit valves tend to be the most inaccurate in their
performance followed by ball and spring valves. The diaphragm valves proved to be most stable in long-term tests.
Most valves, like the slit valves, ballspring, and diaphragm valves, offer a lower resistance (<2.5 mm Hg/
mL/min) than the normal physiological CSF outflow of
610 mm Hg/mL/min. The standard distal tubing of 110
cm increases the overall resistance to 5080% of the physiological value (60).
Standard differential pressure valves are available in
different pressure ranges. It is unclear whether it makes a
difference in an ambulatory patient to use a low-, medium-,
or high-pressure valve because in the upright position
irrespective of the rating the hydrostatic column converts
all differential pressure valves into negative pressure
valves (61). The overdrainage results in persistent headaches from low ICP, ventricular collapse, and increased
risk of shunt obstruction. Long-term changes in cerebrovenous physiology cause acute and severe increase in ICP
without enlargement of ventricles at the time of shunt
malfunction (62). To circumvent the overdrainage in the
11
Figure 14. Three major types of valve designs are available. (a) Differential pressure valves allow
flow in proportion to the pressure difference between the proximal catheter and the distal catheter.
Configurations are a simple diaphragm, ball and spring, or slit valve. Programmable differential
pressure valves can be programmed to a pressure setting using a magnetic wand. In the upright
position, due to a negative pressure from the hydrostatic column of fluid in the distal catheter, these
valves tend to overdrain causing negative pressure symptoms. (b) Flow control valves have the ability
to limit overdrainage from a negative hydrostatic pressure gradient. The Orbis-Sigma Valve has a
wedge-shaped pin over which the orifice of the diaphragm (arrows) rests. When the distal pressure
becomes negative in the upright position, the diaphragm slides downward on the pin narrowing the
drainage channel and hence reducing the flow rate. The Diamond valve has a wedge-shaped slit in
the construct (arrows) that narrows as the distal pressure becomes increasingly negative again
reducing the flow rate. (c) Gravity actuated devices reduce drainage in the upright position by increase
in resistance to flow of CSF from the weight of metal balls in the drainage channel.
upright position, ingenious devices, also referred to as

devices for reducing siphoning (DRS), have been developed
(Fig. 14). The Anti-Siphon device (Integra LifeSciences)
has a flexible diaphragm that mechanically senses atmospheric pressure and shuts off the drainage channel if the
hydrostatic pressure in the fluid column becomes negative.
Flow control valves (Orbis Sigma Valve, Integra LifeSciences and Diamond Valve, Vygon Neuro, Valley Forge,
PA) have a drainage channel that narrows as the differential pressure increases in the upright position to reduce
the flow. Gravity actuated devices (Gravity Compensating
Accessory, Integra LifeSciences, CA, Chabbra Shunt) have
metal balls that fall over one another in the upright position to increase resistance to flow. Double channel devices
(Dual Switch Valve, Christoph Miethke GmBH & Co KG;
SiphonGuard, Codman) have two channels; the low-resistance channel is shut off by a gravity actuated ball in the
upright position.
There is no evidence to suggest that use of one type of
valve is superior to the other, and several valve designs are
available in the market today. A recent multicentric study,
evaluating three basis types of valves, failed to confirm the
utility of flow control or anti-siphon valves in children and
infants over the differential pressure valves (10). Similarly,
studies have failed to show that programmable devices are
superior to fixed pressure valves (63). Over a period of time,
the ventricle tended to become small irrespective of the
type of valve used. The rate of proximal malfunction in a
patient with flow control valves was 6.5% compared with
4246% for the other two valves, although the overall rate
of malfunction and shunt survival was not statistically
different. The design of the flow control valves with a

narrow orifice makes it sensitive to malfunction (64). Certainly, revising a valve has less morbidity and risk of
neurological injury than revising the proximal catheter,
especially in patients with slit ventricles. There is evidence
that a significant number of patients do not tolerate flow
control valves and, despite a radiologically functioning
shunt, have high intracranial pressure from underdrainage through the valve. In patients with limited pressurevolume compensatory reserve, there can be an excessive increase in intracranial pressure during cardiovascular
fluctuations, especially at night and be responsible for nighttime or early morning headaches, in patients with flow
control devices (60). Self-adjusting diaphragm valves like
the Orbis-Sigma (Integra LifeSciences), on bench test, have
proved to be inaccurate and unstable at perfusion rates of
2030 mL/h, which is the most important physiological
range, leading to pre-valve pressures rapidly changing
between 4 and 28 mm Hg. During long-term perfusion,
these may resemble ICP pressure waves (60).
Diaphragm-based anti-siphon devices are prone to
obstruction from encapsulation as has been shown in
experimental animals and is often encountered in patients
who have had recurrent malfunctions (65). Some patients
are more prone to develop heavy scarring around the shunt
system. Again, there is no evidence that using an open
(ASD, Anti Siphon device, Integra LifeSciences) has any
advantage over using a closed system that opens when the
pressure exceeds the negative hydrostatic pressure (SCD,
Siphon Control Device, Medtronics), although theoretically
malfunctions in an open system would only result in loss of
anti-siphon function without obstruction to the flow of

CSF. In the open system (ASD), the flow through the valve
stops only after the intracranial pressure has become
negative in the upright position, which is more physiological, than with SCD, in which the flow stops once the
pressure reaches zero. In the multicentric shunt study, the
incidence of overdrainage was 7.8% in the SCD group and
2.6% in the Standard valve group. The study suggests that
diaphragm-based anti-siphon devices may not be any superior to differential pressure valves in reducing overdrainage
(10). Considerable controversy also revolves around the most
optimum site for placement for the anti-siphon devices
(66,67). The classic position is at the level of the skull base:
however, the bench test suggests a marked tendency to overdrain if the SCD is below the level of the proximal catheter.
These factors may be minor when considered in light of the
excessive sensitivity of the SCD to external pressure from
scar or when the patient is lying on the device (64).
The gravity actuated device (GAD) is used in conjunction
with a differential pressure valve to limit overdrainage (68).
It is similar to the horizontal vertical valve used in lumbar
shunts but constructed to fit in-line with a ventriculoperitoneal shunt. There is no literature to prove or disprove
its utility; however, in individual cases, we have found it
effective. Experimental evidence suggests that motion and
vibration (35) make the mechanism of these devices ineffective although clinical studies are lacking. The position of the
GAD device is critical for optimum functioning. Slight
angulation of the device to vertical can cause underdrainage
in the horizontal position and overdrainage in the vertical
position. Examples of pressure flow characteristics of a
standard differential pressure valve, a flow control valve,
and a valve containing a GAD are shown in Fig. 15.
Distal Catheter
Distal shunt malfunction is reported to occur in 12% to 34%
of shunts (51,69). Three types of distal catheters have been
used: the closed ended with side slits, open ended with side
slits, and open ended. A higher incidence of distal catheter
obstruction has been noted in catheters with side slits
whether closed ended or open ended (51,70). Omental ingrowth is responsible for the peritoneal catheter obstruc-
300
200
High
Medium
100
Low
Pressure (mm H2O)
Pressure (mm H2O)
Shunt Material
Ideal shunt material should be completely biocompatible,
be easy to handle, flexible, resistant to infection, and nonmetallic but radio-opaque (metals interfere with MRI imaging). From a manufacturing standpoint, it should be easy
Low resistance
(safety Valve)
400
400
tion; possibly the distal slits act as collection points for the
debris and provide a channel for trapping the omentum. It
is unclear whether using open-ended distal catheters
increases the likelihood of small ventricle malfunction.
Use of extended length catheters (110120 cm) is not
associated with an increase in the complications and eliminates the need to lengthen the peritoneal catheter for
growth of the patient (71). However, care must be taken
to identify patients who may have enough length of tubing
in the abdomen but may underdrain due to a narrow and
taught segment of tubing from subcutaneous tethering as a
result of scarring and calcification.
It is difficulty to justify use of atrial over the peritoneal
site for distal absorption (72,73). Data on 887 patients
suggested that atrial shunts have a higher rate of malfunction although some studies have not shown a significant difference. However, when the same information was
stratified by age, shunt type, and time period, there was no
significant difference in shunt durability. Cardio-pulmonary
complication, such as irreversible pulmonary hypertension, endocarditis, and glomerulonephritis, are some of
the more serious complications that may occur with atrial
shunts (73). Alternative sites, like pleura, may result in
significant negative pressures in the shunt system (74).
Poor absorption from the pleura may result in large pleural
effusions in small children (74). The gall bladder has also
been effectively used in patients in whom peritoneal, atrial,
or pleural sites have been exhausted (75,76). Potential
complications of these shunts, notably biliary ventriculitis
and biliary meningitis, have been reported in the literature
(77,78). The ventriculo-femoral shunt may be tried in
patients with a difficult access to the atrium from the
subclavian or jugular route (79). Trans-diaphragmatic placement of the distal catheter in the sub-hepatic space
worked successfully in one reported patient with poor
peritoneal access due to scarring (80).
300
High resistance
200
100
Low resistance
10
20 30 40
Flow (cc/hr)
50
60
4 Balls vertical
300
3 Balls vertical
200
2 Balls vertical
100
All valves
horizontal
0
0
400
Pressure (mm H2O)
12
10
20 30 40
Flow (cc/hr)
50
60
10
20
30
Flow (cc/hr)
40
Figure 15. Pressure flow characteristics of (a). Standard differential pressure valve; note that
with increasing differential pressure, such as from upright posture, there is an increase in the flow
rate. (b) The flow control valve has a sigmoid flow-pressure relationship; in the upright position, the
valve works at the high resistance stage and maintains a relatively steady flow rate despite increase
in differential pressure. (c) Gravity actuated device, in vertical position acts as a very highresistance differential pressure valve (depending on the number of balls in the device) and as a
low-resistance differential pressure valve in the supine position.
50
60
to mold into tubing and making valve components. Silicone

polymer is probably the best available material for this
purpose.
Some studies have suggested development of silicone
allergy in some patients with ventricular shunts (8184). It
is unclear whether it represents a true immunological
reaction or a nonspecific foreign body type granulomatous
reaction (85). In patients with suspected or documented
silicone allergy, use of polyurethane (86) or, more recently
CO2 extracted silicone catheters has been postulated but
not proven to offer some advantage in reducing risk of
recurrent malfunctions.
Subcutaneous location of the distal catheters makes
them susceptible to degradation from a foreign-body reaction mounted by the body (87). Scarring around the catheter,
calcification, and stress fractures are long-term consequences of this reaction (88,89). Unless there is some
amount of surface degradation, the adhesions to the subcutaneous tissues do not occur (87). Evidence suggests that
barium used in the silicone catheters is probably not an
important factor in promoting calcification and degradation
(90). Use of barium-free catheters, however, makes it difficult to evaluate a shunt system on radiological imaging.
To minimize colonization of shunt catheters and infection, recently antibiotic-coated catheters have become available. The catheters are available coated with rifampin and
minocycline (Medtronics, Goleta, CA) and another with
rifampin and clindamycin (Bactiseal, Codman, Raynham,
MA). The antibiotic is most active against Staph epidermidis, which is the cause of shunt infection in most patients.
The antibiotic gradually leaches out of the catheter over a
3060 day period providing added advantage. Control studies have shown a significant reduction in rate of infection
with use of these catheters (91,92). The major drawback is
the excessive cost of the antibiotic-coated catheters.
Shunt Malfunction
About 3040% of the shunts malfunction within the first
year of placement (10) and 80% of malfunctions are proximal malfunctions. Although most patients with a malfunctioning shunt will present with the classic features of
raised pressure, headache, and vomiting, in 20%, there
may be no signs of raised pressure (93). Instead, this group
of patients present with a subtle change in behavior,
decline in school performance, gait disturbances, and
incontinence. Some patients may present with aggravation
in the signs and symptoms of Chiari malformation or
syringomyelia. Parents are often more sensitive to these
subtle changes. In a study comparing the accuracy of
referral source in diagnosing shunt malfunction, parents
were more likely to be correct about the diagnosis as
compared with a hospital or general practitioner (94).
At examination, a tense fontanelle, split sutures, and
swelling at the shunt site are very strongly suggestive of a
malfunctioning shunt. Shunt pumping has a positive predictive value of only 20% (95). A shunt valve that fails to fill
up in 10 minutes is very strongly suggestive of shunt
malfunction. Radiological assessment may demonstrate a
fracture or dislocation. Presence of double-backing of the
distal catheter, wherein the distal catheter tip loops out of
13
the peritoneal through the same spot that it enters it, is

diagnostic of distal malfunction (96). The shunt tap gives
useful information about the proximal and distal shunt
system. The absence of spontaneous flow and poor drip rate
indicate proximal malfunction, whereas a high opening
pressure is suggestive of distal malfunction (97). The presence of increase in size of the ventricles on CT scan
confirms a malfunctioning shunt; however, a large number
of patients with long-standing shunt have altered brain
compliance and may not dilate the ventricles at the time of
presentation. In children, similar symptoms occur in the
common illnesses like otitis media; gastroenteritis of viral
fevers often confound the diagnosis. Radiological assessment of shunt flow using radionuclide or iodide contrast
media injected into the shunt may help (2,98100). Unfortunately, although some studies have shown an accuracy of
99% with combined pressure and radionuclide evaluation
(98,101), others have shown a 2540% incidence of deceptive patency when evaluated by radionuclide cisternogram
(97). This could stem from a partial but inadequately
functioning shunt, intermittent malfunction, or presence
of isolated ventricle. Similar problems are encountered
with an iodide contrast-based shuntogram or shunt injection tests. In the absence of normative data with regard to
adequate flow in the shunt, which may vary significantly
with the individual, time of the day, and activity (102), use
of Doppler-based flow devices, flow systems that work
based on differential temperature-gradient or MRI-based
flow systems becomes irrelevant for an individual patient.
Lumbar infusion tests and shunt infusion tests to assess
the outflow resistance through the shunt are cumbersome
and require a laboratory-based setup and may not be
possible in an ER setting (103105). Infusion through a
reservoir to assess outflow resistance through the shunt
suggests a cutoff of less than 12 mm Hg/mL/min as reliable
for distinguishing a clinically suspected high probability of
malfunction from those with a low probability of shunt
malfunction (104). However, this is the group of patients
who may not really need the test, and patients who have a
questionable malfunction on clinical grounds often have
equivocal results on the infusion study.
In childhood hydrocephalus, ICP is the only accurate
guide to shunt function other than the symptoms (105).
Again the ability to measure ICP through the valve tap
becomes unreliable with a partial proximal malfunction. A
similar problem may be encountered with in-line telemetric ICP monitors (106,107). In-addition, the telemetric
transducers may develop a significant drift over time. In
difficult cases, the only way to resolve the issue may be to
explore the shunt, to measure ICP through a lumbar
puncture if the patient has communicating hydrocephalus,
or to place an ICP monitor. Noninvasive monitoring of ICP
is going to have a major role in assessment of these
patients. For patients who have a very compliant brain,
ventricular diltation on the CT scan easily confirms inadequate shunt function.
Despite advances in shunt technology, the incidence of
shunt malfunction has not changed over the last 50 years.
Nulsen and Becker (3) reported a rate of malfunction of
44%, in 1967, which is similar to that reported in recent
studies. To improve on the existing shunt systems, it is
14
Figure 16. Third ventriculostomy is useful

in patients with obstructive hydrocephalus
such as observed with aqueductal stenosis
(arrows). An endoscope is passed through a
small hole in the skull into the frontal horn of
the ventricle and navigated into the third
ventricle. The floor of the third ventricle is
then perforated under vision (arrow, in the
inset) so as to bypass the obstruction at the
aqueduct. Iohexol, a nonionic iodinated dye, is
then instilled into the ventricle to demonstrate
good communication between the ventricles
and the subarachnoid space.
important to understand the factors that reduce shunt

survival. Although the location of the proximal catheter
has not been clearly shown to influence shunt survival
(50,51), the presence of a small amount of fluid around the
proximal catheter is associated with longer shunt survival.
In a study that looked at shunt over a 11 year period,
statistically significant differences were noted in shunt
survival in patients with tumor versus post-hemorrhagic
and aqueductal stenosis; shunts in infants and the pediatric age group survive shorter than in adults; shunt after
multiple revisions survive shorter, and additional shunts
placed for isolated ventricles have shorter survival
(70,108). Chronic inflammatory changes of granular ependymitis often seen at the time of endoscopic shunt placement in patients with multiple revisions probably
contribute to recurrent malfunction and progressive shortening of the interval between revisions as the number of
surgeries increase (108).
The nature of the valve clearly influences the risk of
proximal catheter malfunction. It is much lower with flow
control valves (10,109). Overdrainage from the differential
pressure valves pulls the choroid plexus toward the proximal catheter and may promote malfunction (55). However,
the increased rate of valve malfunction in flow control
devices balances out this advantage (10).
THIRD VENTRICULOSTOMY
In patients with an obstructive type of hydrocephalus,
third ventriculostomy offers an alternative to shunt. The
procedure involves making an opening in the relatively
thin membrane of the floor of the third ventricle. This is

accomplished by passing an endoscope through the lateral
ventricle and guiding it through the foramen of Munro to
the floor of the third ventricle (Fig. 16). The opening allows
CSF to bypass the obstruction at the level of the aqueduct
or the fourth ventricle and directly enter the subarachnoid
space.
Although third ventriculostomy has been recommended
as a procedure of choice for obstructive hydrocephalus;
data from some prospective studies have failed to show
an improved cure rate (110,111). A retrospective analysis of
ventriculographic versus endoscopic third ventriculostomy
in 213 cases does show the superiority of the endoscopic
procedure over the ventriculographic operation both in
terms of reduced risk and improved survival of the procedure (112). Despite the theoretical advantages, evidence
suggests that third ventriculostomy may not be effective in
controlling raised intracranial pressure in all patients
(112,113). Early failures in a radiologically proven case
of obstructive hydrocephalus may relate to multifactorial
etiology of hydrocephalus; associated absorption defects,
obliteration of subarachnoid space from long-standing ventricular diltation, and unidentified infectious cause of aqeductal stenosis may be responsible. Late failures may
relate to gliotic scarring over the ventriculostomy, which
has been visually confirmed by endoscope in some cases.
Does the ventriculostomy close from scarring, or is it a
secondary response to lack of flow through the ventriculostomy due to poor absorption, therefore, a lack of gradient
between the ventricle and the subarachnoid space, is
unclear. In a small prospective study comparing the shunt
failure rate with the failure rate of third ventriculostomy,
no statistical difference was found between the two (12).

Likewise, no controlled study has compared laser, blunt, or
sharp fenestration of the floor or demonstrated usefulness
of balloon diltation of the fenestration. The success rate of
third ventriculostomy of 49100%, as reported in the literature, may not be a true representative of the efficacy of third
ventriculostomy (12). Evaluation after third ventriculostomy
and defining success is difficult in the absence of documented
reduction in ICP or improvement on neuropyschological
tests. This is so because the ventricles may not reduce in
size, and to and from motion through the patent fenestration
may still be observed on MRI CSF flow studies even though
the patient may be symptomatic. In the absence of clear
evidence in literature, third ventriculostomy is often advocated for patients with obstructive hydrocephalus, and if
failures occur, shunting is preferred over repeat fenestration.
It is hoped that close collaboration between the industry
and medicine will help develop smart shunts that would
be able to mimic physiological CSF dynamics. These devices
will possibly incorporate nanotechnology and would be
superior to the presently available devices. It is likely that
better understanding of CSF drainage mechanism in the
future may help develop alternatives such as drugs that
improve drainage of CSF through lymphatic/arachnoidal
venous channels or promote proliferation of new lymphatic/arachnoidal venous channels. Until that time, it seems
that shunts are the best available alternative for management of communicating hydrocephalus.
BIBLIOGRAPHY
1. Dandy WE, Blackfan KD. Internal hydrocephalus: An experimental, clinical and pathological study. Am J Dis Child
1914;30:406482.
2. Graham P, Howman-Giles R, Johnston I, Besser M. Evaluation of CSF shunt patency by means of technetium-99m DTPA.
J Neurosurg 1982;57:262266.
3. Nulsen FE, Becker DP. Control of hydrocephalus by valveregulated shunt. J Neurosurg 1967;26:362374.
4. Pudenz RH, Russell FE, Hurd AH, Shelden CH. Ventriculoauriculostomy; a technique for shunting cerebrospinal fluid
into the right auricle; preliminary report. J Neurosurg
1957;14:171179.
5. Nulsen FE, Spitz EB. Treatment of hydrocephalus by direct
shunt from ventricle to jugular vain. Surg Forum 1951;94:
399403.
6. Fernell E, Hagberg B, Hagberg G, von Wendt L. Epidemiology
of infantile hydrocephalus in Sweden. I. Birth prevalence and
general data. Acta Paediatr Scand 1986;75:975981.
7. Bondurant CP, Jimenez DF. Epidemiology of cerebrospinal
fluid shunting. Pediatr Neurosurg 1995;23:254258; discussion 259.
8. Foltz EL, Shurtleff DB. Five-year comparative study of hydrocephalus in children with and without operation (113 Cases). J
Neurosurg 1963;20:10641079.
9. Laurence KM, Coates S. The natural history of hydrocephalus.
Detailed analysis of 182 unoperated cases. Arch Dis Child
1962;37:345362.
10. Drake JM, Kestle JR, Milner R, Cinalli G, Boop F, Piatt J, Jr.
Haines S, Schiff SJ, Cochrane DD, Steinbok P, MacNeil N.
Randomized trial of cerebrospinal fluid shunt valve design
in pediatric hydrocephalus. Neurosurgery 1998;43:294303;
discussion 303295.
15
11. Patwardhan RV, Nanda A. Implanted ventricular shunts in

the United States: The billion-dollar-a-year cost of hydrocephalus treatment. Neurosurgery 2005;56:139144; discussion
144135.
12. Tuli S, Alshail E, Drake J. Third ventriculostomy versus
cerebrospinal fluid shunt as a first procedure in pediatric
hydrocephalus. Pediatr Neurosurg 1999;30:1115.
13. Bell WO. Cerebrospinal fluid reabsorption. A critical appraisal. 1990. Pediatr Neurosurg 1995;23:4253.
14. Edsbagge M, Tisell M, Jacobsson L, Wikkelso C. Spinal CSF
absorption in healthy individuals. Am J Physiol Regul Integr
Comp Physiol 2004;287:R14501455.
15. Koh L, Zakharov A, Johnston MG. Integration of the subarachnoid space and lymphatics: Is it time to embrace a new
concept of cerebrospinal fluid absorption? Cerebrospinal Fluid
Res 2005;2:6.
16. Johnston M, Zakharov A, Papaiconomou C, Salmasi G, Armstrong D. Evidence of connections between cerebrospinal fluid
and nasal lymphatic vessels in humans, non-human primates
and other mammalian species. Cerebrospinal Fluid Res
2004;1:2.
17. Greitz D, Greitz T, Hindmarsh T. We need a new understanding of the reabsorption of cerebrospinal fluidII. Acta Paediatr 1997;86:1148.
18. Marmarou A, Shulman K, LaMorgese J. Compartmental analysis of compliance and outflow resistance of the cerebrospinal
fluid system. J Neurosurg 1975;43:523534.
19. Raksin PB, Alperin N, Sivaramakrishnan A, Surapaneni S,
Lichtor T. Noninvasive intracranial compliance and pressure
based on dynamic magnetic resonance imaging of blood flow and
cerebrospinal fluid flow: Review of principles, implementation,
and other noninvasive approaches. Neurosurg Focus 2003;14:e4.
20. McAllister JP2nd, Chovan P. Neonatal hydrocephalus.
Mechanisms and consequences. Neurosurg Clin N Am 1998;
9:7393.
21. Adams RD, Fisher CM, Hakim S, Ojemann RG, Sweet WH.
Symptomatic occult hydrocephalus with normal cerebrospinalfluid pressure. A treatable syndrome. N Engl J Med 1965;
273:117126.
22. Oi S, Shimoda M, Shibata M, Honda Y, Togo K, Shinoda M,
Tsugane R, Sato O. Pathophysiology of long-standing overt
ventriculomegaly in adults. J Neurosurg 2000;92:933940.
23. Di Rocco C, Pettorossi VE, Caldarelli M, Mancinelli R, Velardi
F. Experimental hydrocephalus following mechanical
increment of intraventricular pulse pressure. Experientia
1977;33:14701472.
24. Egnor M, Rosiello A, Zheng L. A model of intracranial pulsations. Pediatr Neurosurg 2001;35:284298.
25. Egnor M, Zheng L, Rosiello A, Gutman F, Davis R. A model of
pulsations in communicating hydrocephalus. Pediatr Neurosurg 2002;36:281303.
26. Linninger AA, Tsakiris C, Zhu DC, Xenos M, Roycewicz P,
Danziger Z, Penn R. Pulsatile cerebrospinal fluid dynamics in
the human brain. IEEE Trans Biomed Eng 2005;52:557565.
27. Stephensen H, Tisell M, Wikkelso C. There is no transmantle
pressure gradient in communicating or noncommunicating
hydrocephalus. Neurosurgery 2002;50:763771; discussion
771763.
28. Portnoy HD, Branch C, Castro ME. The relationship of intracranial venous pressure to hydrocephalus. Childs Nerv Syst
1994;10:2935.
29. Marmarou A. A theoretical model and experimental evaluation of the cerebrospinal fluid system. 1973.
30. Alperin N, Lichtor T, Mazda M, Lee SH. From cerebrospinal
fluid pulsation to noninvasive intracranial compliance and
pressure measured by MRI flow studies. Curr Med Imaging
Rev. In press.
16
31. Marmarou A, Shulman K, Rosende RM. A nonlinear analysis

of the cerebrospinal fluid system and intracranial pressure
dynamics. J Neurosurg 1978;48:332344.
32. Ekstedt J. CSF hydrodynamic studies in man. 1. Method of
constant pressure CSF infusion. J Neurol Neurosurg Psych
1977;40:105119.
33. Lundkvist B, Eklund A, Kristensen B, Fagerlund M, Koskinen
LO, Malm J. Cerebrospinal fluid hydrodynamics after placement of a shunt with an antisiphon device: A long-term study.
J Neurosurg 2001;94:750756.
34. Czosnyka M, Batorski L, Laniewski P, Maksymowicz W,
Koszewski W, Zaworski W. A computer system for the identification of the cerebrospinal compensatory model. Acta Neurochirurgica 1990;105:112116.
35. Borgesen SE, Gjerris F. The predictive value of conductance to
outflow of CSF in normal pressure hydrocephalus. J Neurol
1982;105:6586.
36. Marmarou A, Bergsneider M, Klinge P, Relkin N, Black PM.
The value of supplemental prognostic tests for the preoperative assessment of idiopathic normal-pressure hydrocephalus.
Neurosurgery 2005;57:1728.
37. Huang TY, Chung HW, Chen MY, Giiang LH, Chin SC, Lee
CS, Chen CY, Liu YJ. Supratentorial cerebrospinal fluid
production rate in healthy adults: Quantification with twodimensional cine phase-contrast MR imaging with high temporal and spatial resolution. Radiology 2004;233:603608.
38. Alperin NJ, Lee SH, Loth F, Raksin PB, Lichtor T. MRIntracranial pressure (ICP): A method to measure intracranial
elastance and pressure noninvasively by means of MR imaging:
Baboon and human study. Radiology 2000;217:877885.
39. Urchuk SN, Plewes DB. MR measurements of pulsatile pressure gradients. J Magn Reson Imaging 1994;4:829836.
40. Bird R, Stewart W, Lightfoot E. Transport Phenomena. New
York: Wiley Sons; 1960.
41. Walters BC, Hoffman HJ, Hendrick EB, Humphreys RP.
Cerebrospinal fluid shunt infection. Influences on initial management and subsequent outcome. J Neurosurg 1984;60: 1014
1021.
42. Pople IK, Ettles D. The role of endoscopic choroid plexus
coagulation in the management of hydrocephalus. Neurosurgery 1995;36:698701; discussion 701692.
43. Weiss MH, Nulsen FE, Kaufman B. Selective radionecrosis of
the choroid plexus for control of experimental hydrocephalus.
J Neurosurg 1972;36:270275.
44. Aoki N. Lumboperitoneal shunt: Clinical applications, complications, and comparison with ventriculoperitoneal shunt.
Neurosurgery 1990;26:9981003; discussion 10031004.
45. Selman WR, Spetzler RF, Wilson CB, Grollmus JW. Percutaneous lumboperitoneal shunt: Review of 130 cases. Neurosurgery 1980;6:255257.
46. Chumas PD, Armstrong DC, Drake JM, Kulkarni AV, Hoffman HJ, Humphreys RP, Rutka JT, Hendrick EB. Tonsillar
herniation: The rule rather than the exception after lumboperitoneal shunting in the pediatric population. J Neurosurg
1993;78:568573.
47. Payner TD, Prenger E, Berger TS, Crone KR. Acquired Chiari
malformations: Incidence, diagnosis, and management. Neurosurgery 1994;34:429434; discussion 434.
48. Ausman JI. Shunts: Which one, and why? Surg Neurol
1998;49:813.
49. Albright AL, Haines SJ, Taylor FH. Function of parietal and
frontal shunts in childhood hydrocephalus. J Neurosurg
1988;69:883886.
50. Bierbrauer KS, Storrs BB, McLone DG, Tomita T, Dauser R. A
prospective, randomized study of shunt function and infections as a function of shunt placement. Pediatr Neurosurg
1990;16:287291.
51. Sainte-Rose C, Piatt JH, Renier D, Pierre-Kahn A, Hirsch JF,

Hoffman HJ, Humphreys RP, Hendrick EB. Mechanical complications in shunts. Pediatr Neurosurg 1991;17:29.
52. Pang D, Grabb PA. Accurate placement of coronal ventricular
catheter using stereotactic coordinate-guided free-hand passage. Technical note. J Neurosurg 1994;80:750755.
53. Yamamoto M, Oka K, Nagasaka S, Tomonaga M. Ventriculoscopeguided ventriculoperitoneal shunt and shunt revision.
Technical note. Acta Neurochir (Wien) 1994;129:8588.
54. Kestle JR, Drake JM, Cochrane DD, Milner R, Walker ML,
Abbott R, 3rd, Boop FA. Lack of benefit of endoscopic ventriculoperitoneal shunt insertion: A multicenter randomized
trial. J Neurosurg 2003;98:284290.
55. Hakim S. Observations on the physiopathology of the CSF
pulse and prevention of ventricular catheter obstruction in
valve shunts. Dev Med Child Neurol Suppl 1969;20:4248.
56. Martinez-Lage JF, Lopez F, Poza M, Hernandez M. Prevention of intraventricular hemorrhage during CSF shunt revisions by means of a flexible coagulating electrode. A
preliminary report. Childs Nerv Syst 1998;14:203206.
57. Steinbok P, Cochrane DD. Removal of adherent ventricular
catheter. Pediatr Neurosurg 1992;18:167168.
58. Whitfield PC, Guazzo EP, Pickard JD. Safe removal of retained
ventricular catheters using intraluminal choroid plexus coagulation. Technical note. J Neurosurg 1995;83: 11011102.
59. Handler MH. A complication in removing a retained ventricular catheter using electrocautery. Pediatr Neurosurg 1996;
25:276.
60. Czosnyka M, Czosnyka Z, Whitehouse H, Pickard JD. Hydrodynamic properties of hydrocephalus shunts: United Kingdom
Shunt Evaluation Laboratory. J Neurol Neurosurg Psych
1997;62:4350.
61. Trost HA. Is there a reasonable differential indication for
different hydrocephalus shunt systems? Childs Nerv Syst
1995;11:189192.
62. Sood S, Kumar CR, Jamous M, Schuhmann MU, Ham SD,
Canady AI. Pathophysiological changes in cerebrovascular
distensibility in patients undergoing chronic shunt therapy.
J Neurosurg 2004;100:447453.
63. Pollack IF, Albright AL, Adelson PD. A randomized, controlled
study of a programmable shunt valve versus a conventional
valve for patients with hydrocephalus. Hakim-Medos Investigator Group. Neurosurgery 1999;45:13991408; discussion
14081311.
64. Aschoff A, Kremer P, Benesch C, Fruh K, Klank A, Kunze S.
Overdrainage and shunt technology. A critical comparison of
programmable, hydrostatic and variable-resistance valves and
flow-reducing devices. Childs Nerv Syst 1995;11:193202.
65. Drake JM, da Silva MC, Rutka JT. Functional obstruction of
an antisiphon device by raised tissue capsule pressure. Neurosurgery 1993;32:137139.
66. Fox JL, Portnoy HD, Shulte RR. Cerebrospinal fluid shunts:
An experimental evaluation of flow rates and pressure values
in the anti-siphon valve. Surg Neurol 1973;1:299302.
67. Tokoro K, Chiba Y. Optimum position for an anti-siphon
device in a cerebrospinal fluid shunt system. Neurosurgery
1991;29:519525.
68. Chabbra DK, Agarwal GD, Mittal P. Z flow hydrocephalus
shunts, a new approach to the problem of hysrocephalus. The
rationale behind its design and the initial results of pressure
monitoring after Z flow shunt implantation. Acta Neurochir
(Wien) 1993;121:4347.
69. Sekhar LN, Moossy J, Guthkelch AN. Malfunctioning ventriculoperitoneal shunts. Clinical and pathological features. J
Neurosurg 1982;56:411416.
70. Cozzens JW, Chandler JP. Increased risk of distal ventriculoperitoneal shunt obstruction associated with slit valves or
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
distal slits in the peritoneal catheter. J Neurosurg 1997;87:

682686.
Couldwell WT, LeMay DR, McComb JG. Experience with use
of extended length peritoneal shunt catheters. J Neurosurg
1996;85:425427.
Borgbjerg BM, Gjerris F, Albeck MJ, Hauerberg J, Borgesen
SV. A comparison between ventriculo-peritoneal and
ventriculo-atrial cerebrospinal fluid shunts in relation to rate
of revision and durability. Acta Neurochir (Wien) 1998;140:
459464; discussion 465.
Lam CH, Villemure JG. Comparison between ventriculoatrial and ventriculoperitoneal shunting in the adult population. Br J Neurosurg 1997;11:4348.
Willison CD, Kopitnik TA, Gustafson R, Kaufman HH. Ventriculopleural shunting used as a temporary diversion. Acta
Neurochir (Wien) 1992;115:6768.
Ketoff JA, Klein RL, Maukkassa KF. Ventricular cholecystic
shunts in children. J Pediatr Surg 1997;32:181183.
Novelli PM, Reigel DH. A closer look at the ventriculogallbladder shunt for the treatment of hydrocephalus.
Pediatr Neurosurg 1997;26:197199.
Barami K, Sood S, Ham SD, Canady AI. Postural changes in
intracranial pressure in chronically shunted patients.
Pediatr Neurosurg 2000;33:6469.
Bernstein RA, Hsueh W. Ventriculocholecystic shunt. A
mortality report. Surg Neurol 1985;23:3137.
Philips MF, Schwartz SB, Soutter AD, Sutton LN. Ventriculofemoroatrial shunt: A viable alternative for the treatment
of hydrocephalus. Technical note. J Neurosurg 1997;86:
10631066.
Rengachary SS. Transdiaphragmatic ventriculoperitoneal
shunting: Technical case report. Neurosurgery 1997;41:
695697; discussion 697698.
Goldblum RM, Pelley RP, ODonell AA, Pyron D, Heggers JP.
Antibodies to silicone elastomers and reactions to ventriculoperitoneal shunts. Lancet 1992;340:510513.
Gower DJ, Lewis JC, Kelly DL, Jr. Sterile shunt malfunction.
A scanning electron microscopic perspective. J Neurosurg
1984;61:10791084.
Snow RB, Kossovsky N. Hypersensitivity reaction associated
with sterile ventriculoperitoneal shunt malfunction. Surg
Neurol 1989;31:209214.
Sugar O, Bailey OT. Subcutaneous reaction to silicone in
ventriculoperitoneal shunts. Long-term results. J Neurosurg
1974;41:367371.
Kalousdian S, Karlan MS, Williams MA. Silicone elastomer
cerebrospinal fluid shunt systems. Council on Scientific
Affairs, American Medical Association. Neurosurgery 1998;
42:887892.
Jimenez DF, Keating R, Goodrich JT. Silicone allergy in
ventriculoperitoneal shunts. Childs Nerv Syst 1994;10: 59
63.
Del Bigio MR. Biological reactions to cerebrospinal fluid
shunt devices: A review of the cellular pathology. Neurosurgery 1998;42:319325; discussion 325316.
Echizenya K, Satoh M, Murai H, Ueno H, Abe H, Komai T.
Mineralization and biodegradation of CSF shunting systems.
J Neurosurg 1987;67 :584591.
Elisevich K, Mattar AG, Cheeseman F. Biodegradation of
distal shunt catheters. Pediatr Neurosurg 1994;21:7176.
Irving IM, Castilla P, Hall EG, Rickham PP: Tissue reaction
to pure and impregnated silastic. J Pediatr Surg 1971;6:724
729.
Aryan HE, Meltzer HS, Park MS, Bennett RL, Jandial R,
Levy ML. Initial experience with antibiotic-impregnated silicone catheters for shunting of cerebrospinal fluid in children.
Childs Nerv Syst 2005;21:5661.
17
92. Govender ST, Nathoo N, van Dellen JR: Evaluation of an

antibiotic-impregnated shunt system for the treatment of
hydrocephalus. J Neurosurg 2003;99:831839.
93. Fried A, Shapiro K. Subtle deterioration in shunted childhood
hydrocephalus. A biomechanical and clinical profile. J Neurosurg 1986;65:211216.
94. Watkins L, Hayward R, Andar U, Harkness W. The diagnosis
of blocked cerebrospinal fluid shunts: A prospective study of
referral to a paediatric neurosurgical unit. Childs Nerv Syst
1994;10:8790.
95. Piatt JH, Jr. Physical examination of patients with cerebrospinal fluid shunts: Is there useful information in pumping the shunt? Pediatrics 1992;89:470473.
96. Martinez-Lage JF, Poza M, Izura V. Retrograde migration of
the abdominal catheter as a complication of ventriculoperitoneal
shunts: The fishhook sign. Childs Nerv Syst 1993;9: 425427.
97. Sood S, Canady AI, Ham SD. Evaluation of shunt malfunction using shunt site reservoir. Pediatr Neurosurg 2000;
32:180186.
98. Hayden PW, Rudd TG, Shurtleff DB. Combined pressureradionuclide evaluation of suspected cerebrospinal fluid
shunt malfunction: A seven-year clinical experience. Pediatrics 1980;66:679684.
99. Sweeney LE, Thomas PS. Contrast examination of cerebrospinal fluid shunt malfunction in infancy and childhood.
Pediatr Radiol 1987;17:177183.
100. Vernet O, Farmer JP, Lambert R, Montes JL. Radionuclide
shuntogram: Adjunct to manage hydrocephalic patients. J
Nucl Med 1996;37:406410.
101. Savoiardo M, Solero CL, Passerini A, Migliavacca F. Determination of cerebrospinal fluid shunt function with watersoluble contrast medium. J Neurosurg 1978;49:398407.
102. Kadowaki C, Hara M, Numoto M, Takeuchi K, Saito I. CSF
shunt physics: factors influencing inshunt CSF flow. Childs
Nerv Syst 1995;11:203206.
103. Czosnyka M, Whitehouse H, Smielewski P, Simac S, Pickard
JD. Testing of cerebrospinal compensatory reserve in
shunted and non-shunted patients: A guide to interpretation
based on an observational study. J Neurol Neurosurg Psych
1996;60:549558.
104. Morgan MK, Johnston IH, Spittaler PJ. A ventricular infusion technique for the evaluation of treated and untreated
hydrocephalus. Neurosurgery 1991;29:832836; discussion
836837.
105. Fouyas IP, Casey AT, Thompson D, Harkness WF, Hayward
RD. Use of intracranial pressure monitoring in the management of childhood hydrocephalus and shunt-related problems. Neurosurgery 1996;38:726731; discussion 731722.
106. Woodford J, Saunders RL, Sachs E, Jr. Shunt system patency
testing by lumbar infusion. J Neurosurg 1976;45:6065.
107. Cosman ER, Zervas NT, Chapman PH, Cosman BJ, Arnold
MA. A telemetric pressure sensor for ventricular shunt systems. Surg Neurol 1979;11:287294.
108. Miyake H, Ohta T, Kajimoto Y, Matsukawa M. A new ventriculoperitoneal shunt with a telemetric intracranial pressure
sensor: Clinical experience in 94 patients with hydrocephalus.
Neurosurgery 1997;40:931935.
109. Lazareff JA, Peacock W, Holly L, Ver Halen J, Wong A,
Olmstead C. Multiple shunt failures: An analysis of relevant
factors. Childs Nerv Syst 1998;14:271275.
110. Decq P, Barat JL, Duplessis E, Leguerinel C, Gendrault P,
Keravel Y. Shunt failure in adult hydrocephalus: Flow-controlled shunt versus differential pressure shuntsa cooperative study in 289 patients. Surg Neurol 1995;43:333339.
111. Garton HJ, Kestle JR, Cochrane DD, Steinbok P. A costeffectiveness analysis of endoscopic third ventriculostomy.
Neurosurgery 2002;51:6977; discussion 7768.
18
HYPERBARIC MEDICINE
112. Santamarta D, Diaz Alvarez A, Goncalves JM, Hernandez J.

Outcome of endoscopic third ventriculostomy. Results from
an unselected series with noncommunicating hydrocephalus.
Acta Neurochir (Wien) 2005;147:377382.
113. Cinalli G, Sainte-Rose C, Chumas P, Zerah M, Brunelle F,
Lot G, Pierre-Kahn A, Renier D. Failure of third ventriculostomy in the treatment of aqueductal stenosis in children. J
Neurosurg 1999;90:448454.
114. Hirsch JF, Hirsch E, Sainte Rose C, Renier D, Pierre-Khan A.
Stenosis of the aqueduct of Sylvius. Etiology and treatment. J
Neurosurg Sci 1986;30:2939.
See also INTRAUTERINE SURGICAL TECHNIQUES; MICRODIALYSIS SAMPLING;
MONITORING, INTRACRANIAL PRESSURE.
HYPERALIMENTATION.
See NUTRITION, PARENTERAL.
HYPERBARIC MEDICINE
BARBARA L. PERSONS
BETH COLLINS
WILLIAM C. LINEAWEAVER
University of Mississippi
Medical Center
Jackson, Mississippi
INTRODUCTION
The goal of hyperbaric oxygen (HBO) therapy is to deliver
high concentrations of oxygen under pressure to increase
the amount of dissolved oxygen in the blood. The physiologic repercussions of this increased plasma oxygen have
widespread effects that translate into a variety of clinical
applications. Initially, the use of hyperbaric medicine surrounded acute decompression illness and gas embolism.
Later, the increased oxygen under pressure was shown to
have use in a variety of clinical situations. The delivered
pressure can be two to six times ambient atmospheric
pressure (ATM) or atmospheres absolute (ATA) depending
on the indication. The current Undersea and Hyperbaric
Medical Society approved uses of HBO are shown in Table 1,
and many of these indications will be discussed individually.
Table 1. Approved Uses for Hyperbaric

Oxygen Therapy
acute decompression illness
gas embolism
carbon monoxide poisioning
clostridial gas gangrene
necrotizing soft tissue infections
compromised skin grafts and skin flaps
crush injury
compartment syndrome
acute traumatic ischemias
radiation tissue damage
refractory osteomyelitis
selected problem wounds
acute exceptional blood loss anemia
acute thermal burns
intracranial abscess
HBO in the treatment of these conditions is supported by

controlled medical trials published in peer-reviewed journals and, as such, is evidence-based. Numerous other
experimental uses exist for HBO, such as for stroke and
for cardiac ischemia, but these uses have not yet been
sufficiently proven to be supported by the Undersea and
Hyperbaric Medicine Society or by the American College
of Hyperbaric Medicine. The goal of this chapter is to
outline the physical principles underlying the use of HBO
therapy, to discuss its medical indications for HBO, and
to familiarize the reader with the mechanical, safety,
and regulatory issues involved in operating a hyperbaric
medicine program.
HISTORICAL BACKGROUND
British physician and clergyman Henshaw was the first to
use alteration in atmospheric pressure to treat medical conditions when he used his domicilium chamber in 1862.
Hyperbaric medicine also surrounded diving and diving
medicine. Triger, in 1841, gave the first human description
of decompression sickness (1). In 1934, U.S. Naval Submarine Officer Dr. Albert Behnke was the first to use oxygen
recompression to treat decompression sickness in naval
divers (2). Later, in 1943, Gagnon and Cousteau invented
SCUBA (self-contained underwater breathing apparatus).
Dr. Boerma, a Dutch thoracic surgeon, removed the blood
cells from pigs in 1955 and found they could survive with the
oxygen dissolved in plasma by use of HBO. An upsurge in
hyperbaric surgery followed in 1956, when Boerma performed cardiovascular surgery in a hyperbaric chamber,
which along with hypothermia, allowed for periods of circulatory arrest of 78 min. The large chamber developed at
Wilhelmina Gasthuis in Amsterdam in 1959, headed by
Boerma, allowed a wide variety of research to be carried
out on the uses of HBO therapy on many diseases (3). By
1966, it was indicated for the treatment of protection during
induced circulatory arrest, homotransplantation, clostridial
infection, acute arterial insufficiency, chronic arterial insufficiency, and hypovolemic shock. Shortly thereafter, the
advent of cardiopulmonary bypass obviated hyperbaric
chambers for cardiac protecton. In 1967, the Undersea Medical Society was founded by the U.S. Navy diving and submarine medical officers. This organization originally focused
on undersea and diving medicine but, later, came to include
clinical hyperbaric medicine. In 1986, the name was changed
to the Undersea and Hyperbaric Medical Society or UHMS
with more than 2500 physician and scientist members in
50 countries. More recently, the American College of Hyperbaric Medicine has come to offer board certification to U.S.
physicians in the specialty of hyperbaric medicine.
PHYSICS
To understand HBO therapy, one must understand a few
basic laws of physics, namely, Boyles law, Charles law,
Daltons law, and Henrys law. Boyles law explains how gas
volume shrinks with increasing pressure. Charles law
explains that the volume of a gas decreases with decreasing
temperature. Daltons law explains that each gas in a
HYPERBARIC MEDICINE
mixture exerts its own partial pressure independently of the

others. Henrys law explains that the number of gas molecules
that will dissolve in a liquid depends on the partial pressure of
the gas as well as on the mass of the liquid. These laws
themselves will be explained in the first part of this section
and the application of the laws to each area of hyperbaric
medicine will be explained in the respective section.
Boyles Law
Boyles law states if the temperature remains constant, the
volume of a gas is inversely proportional to the pressure.
Boyles law is stated as PV K or P K/V, where P is
pressure, V is volume, and K is a constant.
Thus, as in Table 2, the volume of a bubble at 1 Atmosphere or sea level shrinks to one-half of its original volume
at 33 feet below sea level (10 m or 2 atmospheres). It
shrinks to one-third of its original size at 66 feet below
sea level (20 m or 3 atmospheres). Conversely, if a diver
were to hold his breath at depth and then ascend, the air in
his lungs will expand to three times the volume it occupied
at 66 feet below sea level.
Charles Law
Charles law states that if the pressure remains constant,
the volume of a fixed mass of gas is directly proportional to
absolute temperature. The volume increases as the temperature increases. For example, a balloon has a volume of
1 L at 20 8C and its volume would expand to 1.1 L at 50 8C.
Charles law is stated as V/T K or V1/T1 V2/T2,
where V is volume of gas 1 or 2 , T is temperature in Kelvin
of gas 1 or 2, and K is a constant.
Daltons Law
Daltons law states that each gas in a mixture exerts its
partial pressure independently, which is important in understanding human physiogy and HBO. For example, if a patient
is given a high concentration of oxygen and a lower concentration of nitrogen, these gasses will diffuse across membranes and act in the body independently of each other.
Daltons law is stated as P(t) P1P2P3, where P(t)
is total pressure and P1, P2, and P3 are the individual gas
pressures. Gases try to equalize their concentrations across
a membrane, which explains how breathing a higher oxygen, lower nitrogen mixture can help nitrogen leave the
blood through the lungs as nitrogen gas.
Henrys Law
Henrys law states that at a given temperature, the amount
of gas dissolved in a solute is directly proportional to the
pressure of the gas above the substance.
Table 2. Pressure vs. Volume at Depth
Depth
(feet sea water)
0(sea level)
33
66
165
Pressure (ATA)
atmospheres
1
2
3
6
19
Henrys law illustrates that when a liquid is exposed to a

gas, some of the gas molecules dissolve into it. The number
of moles that will dissolve in the liquid depends on the mass
of the liquid, the partial pressure of the gas, its solubility in
the liquid, the surface area of contact, and the temperature
(as that changes with the partial pressure). Thus, more
gas, oxygen, or nitrogen will dissolve in tissue fluid at a
higher pressure because the partial pressure of each gas
increases at a higher pressure.
Henrys law is stated as p Kc, where p is the partial
pressure of the gas 1, c is its molar concentration, and K is
the Henrys law constant, which is temperature-dependent.
An example of Henrys law is dissolved carbon dioxide
in soda, which bubbles out of solution as the pressure
decreases. Another example is exemplified by water when
it is heated. Long before it boils, bubbles of air form on the
side of the pan, which is an example of the gas coming out of
solution as the temperature is raised.
For treatment of decompression illness (DCI) and gas
embolism (GE), increased pressure alone as well as the
increased oxygen pressure facilitate treatment. Both of these
conditions are a result of gas bubbles in the tissues or gas
bubbles in the blood causing blockage of vessles or ischemia of
tissues. Therefore, shrinking the bubbles with increased pressure allows them to be removed or minimized by the body,
which is the principle of Boyles law, that the volume of a gas
varies inversely with pressure. If one has a bubble occluding
an important vessel or lodged in a joint, it will shrink as the
pressure increases as in Table 2. The blood can accommodate
an increased amount of dissolved gas with increased atmospheric pressure as explained by Henrys law. Henrys law
states that the amount of gas that will dissolve in a liquid is
proportional to the partial pressure of the gas in contact with
that liquid as well as to the atmospheric pressure. Atmospheric pressure at sea level is 760 mmHg, and the normal
atmosphere consists of 21% oxygen and 79% nitrogen. (see
Fig. 1) (4). The pressure of a gas dissolved in plasma relates to
its solubility in a liquid as well as to its partial pressure. A gas
bubble caught in the tissues or in the systemic circulation
either because of an air embolus or because of decompression
sickness is significantly decreased under hyperbaric conditions, as illustrated by Table 3. By Boyles law, the pressure
alone shrinks the bubble to a fraction of its original size.
Daltons law states that total pressure of a mixture of gases
is equal to the sum of the pressures of each gas. Each gas is
acting as if it alone were present, which explains how the
oxygen under pressure creates a situation whereby the higher
dissolved oxygen surrounds the bubble and causes the diffusion of nitrogen out of the bubble, called nitrogen washout.
Simply put, each gas attempts to have equal concentration
of particles, in this case, nitrogen and oxygen on each side
of the gas bubble. Nitrogen, therefore, diffuses out of the
bubble and shrinks in size. Hyperbaric oxygen enables
treatment of the two conditions where air or nitrogen
bubbles become lodged in the tissues.
Gas Volume (%)

100
50
33
17
PHYSIOLOGY
Hyperbaric oxygen therapy oxygen enters the systemic
circulation through the alveoli in the lungs, and the
20
HYPERBARIC MEDICINE
Table 3. Oxygen Levels During Hyperbaric Oxygen Treatment Breathing Air (4)
ATA Chamber
Pressure Chamber
PO2 (mmHg) Chamber
PAO2 (mmHg) Lung
O2 ml/dl vol % Plasma
760
1520
1794
2143
2280
3040
3800
4560
160
319
377
450
479
638
798
958
100
269
322
400
429
588
748
908
0.31
0.83
1.00
1.24
1.33
1.82
2.32
2.81
1
2
2.36
2.82
3
4
5
6
Breathing 100% Oxygen
1
2
2.36
2.82
3
4
5
6
760
1520
1794
2143
2280
3040
3800
4560
760
673
2.08
1520
1433
4.44
1794
1707
5.29
2143
2056
5.80
2280
2193
6.80
100% oxygen is not used above 3ATA to minimize the risk of oxygen toxicity.
diffusion is mediated by the pressure differential between

the alveolar oxygen content and the oxygen content of
venous blood. Alveolar oxygen content is 100 mmHg and
venous oxygen content is 40 mm Hg (5). Normally, 97% of
oxygen iscarried in the arterial blood bound to hemoglobin
molecules and only 3% of the oxygen is dissolved in plasma,
illustrated by the formula for oxygen content in arterial
blood. CaO2 (1.34
Hb
SaO2) (0.003
PaO2), where
CaO2 is oxygen content, Hb is Hemoglobin in grams, and
SaO2 is arterial O2 saturation expressed as a fraction not a
percentage (0.95, not 95%). 1.34 is the realistic binding
capacity of hemoglobin, although 1.39 is the actual binding
capacity. 0.003 times the PaO2 is the amount of oxygen
soluble in plasma at normal atmospheric pressure. With
hyperbaric oxygen, the amount of oxygen dissolved in arterial blood is dramatically increased. Conversely, the amount
carried by hemoglobin remains about the same as that
achieved by inspiring oxygen at 1 atmosphere absolute
(ATA) (4). As measured in ml O2 per deciliter of whole blood,
the oxygen content increases significantly under hyperbaric
conditions as shown in Table 3. The oxygen content of blood
increases from 0.31 at 1 atmosphere to 6.80ml/dl vol% at
3 atmospheres. Note that 100% oxygen is not used at
pressures greater than 3 ATA to minimize the risk of oxygen
toxicity. The alveolar type 1 cells in the lungs and the
neurons in the brain are sensitive to excessive concentrations of oxygen. Again, to prevent oxygen toxicity, which can
lead to alveolar damage or seizure, 100% oxygen is not given
at pressures greater than 3 ATA. Even in these pressure
ranges, air breaks are given to patients and they breathe air
as opposed to concentrated oxygen under pressure for 10
minutes during many of the protocols, which in theory, gives
the xanthine oxidese system a chance to deal with the
current load of free radicals in the lungs and tissues and
the Gaba amino buteric acid (GABA) depletion in the brain a
chance to normalize. As mentioned, the bloods ability to
carry more dissolved oxygen molecules under higher atmospheric pressure is a function of Henrys law. It states that
the amount of gas that will dissolve in a liquid is propor-
tional to the partial pressure of the gas in contact with that

liquid as well as to the atmospheric pressure. Thus, more
gas, oxygen, or nitrogen will dissolve in tissue fluid at higher
atmospheric pressure. %X is the percentage of gas X dissolved in the liquid, P(t) is the atmospheric pressure, and P(X)
is the partial pressure of gas X.
TRANSCUTANEOUS OXYMETRY (TcPO2 OR TCOM)
In order for the patient to benefit from HBO therapy, they
must have adequate perfusion of blood to the affected area.
This perfusion can be assessed prior to hyperbaric oxygen
therapy by checking transcutaneous oxygen tension,
TcPO2. Transcutaneous oxygen tension values of less than
40, which increase to more than 100 mmHg while breathing 100% oxygen or to more than 200 during HBO therapy,
will likely benefit from hyperbaric oxygen therapy (6). The
detailed mechanisms by which the elevated oxygen tension
is felt to improve wound healing and the bodys ability to
combat bacterial pathogens will be discussed under each
indication for HBO therapy. Briefly, Hunt and Pa: (7)
showed, in 1976, that increased oxygen tension stimulated
collagen synthesis and fiberblast proliferation. Then, studies revealed improved ability of leukocytes to clear infected
wounds of bacteria with hyperbaric oxygen (8,9). Then, in
1989, Zamboni et al. (10) showed that HBO therapy in the
reduction of tissue flap necrosis was a systemic phenomenon
that involved inhibition of neutrophil adherence and prevention of arteriolar vasoconstriction thought to be via a
nitric oxide-mediated mechanism. In addition, it increases
platelet-derived growth factor Beta (PDGF B), vascular
endothelial growth factor (VEGF), Epidermal growth factor
(EGF), and other factors.
APPROVED INDICATIONS
The following indications are approved uses of hyperbaric
oxygen therapy as defined by the Hyperbaric Oxygen
HYPERBARIC MEDICINE
Therapy Committee of the Undersea and Hyperbaric

Medical Society (Table 1). Most of these indications are
also covered by Medicare and many are covered by major
insurance companies. The indications include air or gas
embolism, decompression illness, carbon monoxide poisioning, Clostridial myositis and myonecrosis (gas gangrene), crush injury, compartment syndrome and other
acute traumatic ischemias, decompression sickness, problem wounds, exceptional blood loss anemia, intracranial
abscess, necrotizing soft tissue infections, refractory osteomyelitis, delayed radiation injury, compromised skin grafts
and flaps, and thermal burns. Many of these indications will
be specifically discussed in the following sections.
21
Other
1%
Oxygen
21%
HYPERBARIC CHAMBER BASICS

Hyperbaric oxygen therapy is a feature offered in many
hospitals, medical centers, and in specialty situations
such as diver rescue stations and oil rigs. Over 500
hyperbaric chambers exist in the United States alone.
When a patient is refered for one of the listed emergent
or nonemergent indications to undergo hyperbaric oxygen
therapy, a complete workup of the patient should be
performed if possible prior to hyperbaric oxygen therapy.
Emergency situations may necessitate an abbreviated
exam. The patient should wear only cotton medical-center-provided clothing to prevent static electricity, which
could cause a spark and a fire. All foreign appliances
should be removed, including hearing aids, lighters,
and jewelry. Internal appliances such as pacemakers
are usually safe under hyperbaric conditions. The patient
will then be premedicated with a benzodiazapine such as
valium if needed for anxiety. Hyperbaric oxygen therapy
can be administered through monoplace or multiplace
chambers. Monoplace chambers accommodate a single
patient within a pressurized environment of 100% oxygen
(Fig. 1). These chambers are often constructed as acrylic
cylinders with steel ends and can withstand pressures of
up to 3 ATA. Multiplace chambers are usually constructed
of steel and can withstand pressures up to 6 ATA (Fig. 2).
They can accommodate two or more people and often have
the capacity to treat ventilated or critically ill patients.
Some are even large enough to accommodate operating
teams, and the 100% oxygen is delivered to the patient via
face mask, hood, or endotrachial tube. Depending on the
indication, patients will require from 1 to 60 treatments.
Most commonly, treatments are delivered to the patient
for 6090 min at 2.83.0 ATA five days a week for the
protocol duration.
Nitrogen
78%
Figure 1. Relative composition of air.
3. Bleomycin History of the chemotherapy agent

Bleomycin, which can cause pneumonitis, especially
if the patient is exposed to HBO.
4. Doxyrubicin.
5. Disulfiram (antibuse) Blocks production of superoxide dismutase, which protects the patient from
oxygen toxicity.
6. Cisplatin/Carboplatin An anticancer agent that
interferes with DNA synthesis.
7. Mefanide (Sulfamyelon) It is a topical ointment for
burns and wounds that is a carbonic anhydrase
inhibitor and increases the risk of seizure during
HBO therapy.
The relative considerations and contraindications are
many. In these patients, the hyperbaric physician should
consider each patient individually, including their history
of thoracic surgery, seizure disorder, obstructive lung disease, congestive heart failure, pulmonary lesions on X-ray
or CT scan, seizure disorder, upper respiratory infections
CONTRAINDICATIONS
Six absolute contraindications exist to hyperbaric oxygen
therapy.
1. Untreated pneumothorax An untreated pneumothorax can be converted to a tension pneumothorax with administration of HBO.
2. History of spontaneous pneumothorax.
Figure 2. Fire risk.
22
HYPERBARIC MEDICINE
Table 4. Arterial and Venous Air Embolism Etiology

Etiology of Air Embolism
Arterial Air Embolism
Pulmonary Overpressure
Open Lung Biobsy
Arterial Catheter
Angiography
Surgery
Penetrating Chest Trauma
Pneumothorax
Cardiopulmonary Bypass
Dialysis
Autotransfusion
Neonatal Respiratory
Distress Syndrome
Paradoxical, Patent
Foramen Ovale
Venous Air Embolism

Central Venous Catheters
Infusion Pumps
Neurosurgery
Laparoscopy
Liver Transplantation
Neurosurgery
Pelvic Surgery
Trendelenburg Position
Necrotizing Enterocolitis
Lumbar Spine Spine Surgery
Air Contrast Salpingogram
Umbilical Venous Catheters
(due to risk of barotraumas to the ears), acute viral infections, uncontrolled high fever (as it increases CNS sensitivity to oxygen), reconstructive ear surgery, congenital
spherocytosis, history of optic neuritis, recent retinal
repair, claustrophobia, acidosis, nicotine, alcohol, and
many others.
AIR EMBOLISM
Air embolism is a medical emergency. In the diving community, it is the main cause of death following diving
accidents. Early diagnosis followed by definitive treatment are critical in determining the eventual outcome.
Treatment is based on compression of the air bubbles by
Boyles law as well as oxygenation of ischemic tissues and
treatment of ischemia reperfusion injury with hyperbaric
oxygen. Air embolism can occur in the hospital setting
by introduction of air into the systemic circulation by
central venous and arterial catheters and other invasive
procedures. Interestingly, the first report of a death from
air embolism was from France in 1821. The patient was
undergoing surgery on his clavicle when the surgeon
noted a hissing sound in the area of operation and the
patient yelled my blood is falling into my heart- Im
dead (11). The patient likely died of a venous air embo-
lism obstructing the systemic circulation. Air entry in

the systemic circulation occurs following violation of
the systemic circulation by any number of mechanisms
(Table 4), which can be either by introduction of air into
the arterial circulation, as in a lung biopsy, chest trauma,
and pulmonary overpressure (diving), or into venous circulation, as in air introduction via central venous catheters, liver transplantation, and neurosurgery. Venous air
emboli are more common, whereas arterial emboli are tend
to be more serious. Physiologically, the air bubble forms or is
introduced into the circulation. The lung usually serves as
an excellent filter for air emboli, and can protect the embolism from traveling to the brain. This protective filter may be
bypassed by a patent foramen ovale. Approximately 30% of
patients have a foramen ovale that is patent by probe. The
lung as a filter may be overwhelmed by large quantities of
air. The bubble can then lodge in the smaller arteries of the
brain causing obstruction. An air embolism is immediately
identified as a foreign body, and platelets are activated,
which leads to an inflammatory cascade. Hypoxia then
develops distal to the obstruction with associated swelling.
The embolism is eventually absorbed by the body, but the
fibrin deposition at the embolism site may prevent return of
blood flow. In order to diagnose an air embolism, it needs to
have a high index of suspicion. If air embolism is suspected,
the patient should receive a number of immediate measures,
including ACLS or ATLS protocols. The patient should be
placed on the left side and one should consider draining air
from the right atrium with a central venous catheter. Air in
the heart can cause a machinery murmur. 100% oxygen
should be administered via a face mask or endotrachial tube,
and the patient should be hydrated to preserve intravascular volume. Per Daltons law, administration of high
oxygen will cause nitrogen to diffuse out of the air bubble
and shrink in size. Daltons law states that total pressure of
a mixture of gases is equal to the sum of the pressures of
each gas. Each gas acts as if it alone were present. Daltons
law is stated as P(t) P1 P2 P3. The inspired 100%
oxygen will also maximize oxygenation of the tissues as
much as possible under normal atmospheric pressure. Next,
the patient should be emergently treated with hyperbaric
oxygen. If a chamber is available that can provide compression to 6 ATA with air or a mixture of 50% nitrogen and 50%
air, treatment should immediately be performed following
the U.S. Navy protocol (Fig. 3).
Figure 3. U.S. Navy decompression treatment.
HYPERBARIC MEDICINE
DECOMPRESSION ILLNESS
By definition, decompression illness (DCI), also called
bends or caisson disease, occurs when gas bubbles exit
in the blood or body tissues. At depth, more gas can
dissolve in the tissues than in the blood. When the diver
ascends too quickly, the gas comes out of solution and
forms bubbles in the tissues and in the blood, much like
popping a soda can and releasing its pressure causes the
carbon dioxide bubbles to come out of the solution. In
1994, Divers Alert Network (DAN) recorded 1164 divingrelated injuries and 97 diving-related deaths, many
related to DCI (12). The severity of DCI depends on the
volume and location of gas bubbles. The range of symptoms is from vague constitutional complaints or limb pain
to cardiopulmonary arrest and coma, the pathophysiology
of which is explained by Henrys law. As previously
explained, Henrys law is p Kc, where p is the partial
pressure of the gas 1, c is its molar concentration, and K is
the Henrys law constant, which is temperature-dependent. Thus, simply stated, more inert gas can be dissolved
in a liquid at higher atmospheric pressure and, conversely, less under lower atmospheric pressure, which occurs
on decompression when gas must be removed from tissues, and rapid decompression leads to bubble formation.
Using the Henrys law equation, one can calculate the
estimated amount of nitrogen a diver must clear from the
bloodstream (about 5 L) in rising from 100 ft to the surface. The amout would be approximately 750 ml nitrogen
assuming room temperature, which is a significant
volume of nitrogen that must be eliminated from the
divers bloodstream. The onset of symptoms is usually
rapid and 75% of patients experience symptoms within
1 h of decompression and 90% within 12 h of decompression (13). A small number of patients may present even
later, particularly if they have flown in commercial aircraft after diving and not followed the recommendation of
the major diving organizations not to fly within 24 hours
of ones last dive. Interestingly, up to 10% of the inert gas
that is absorbed in the tissues is released as bubbles after
the divers decompression (14). Patients experience symptoms depending on the location and concentration of the
Table 5. Signs and Symptoms of Decompression Illness

Symptoms of
Decompression Illness
Signs of
Decompression Illness
Unusual fatigue
Skin itching
Pain in joints/
muscles of arms,
legs or torso
Dizziness
Blotchy skin rash

Paralysis, muscle weakness
Difficulty urinating
Vertigo
Ringing in the ears, (Tinnitus)
Numbness, tingling
and paralysis
Shortness of breath (Dyspnea)
Confusion, personality
changes, bizarre behavior
Amnesia,
Staggering
Coughing up bloody,
frothy sputum
Collapse or unconsciousness
Tremors
23
bubbles (see Table 5). Bubbles forming in or near joints

cause the joint pain of a classical bend. These musculoskeletal effects are called type 1 DCS. When these
effects occur in the spinal cord or brain, numbness,
paralysis, and disorders of higher cerebral function may
result. If large numbers of bubbles enter the venous
bloodstream, congestive symptoms in the lung and circulatory shock can then occur. These pulmonary and neurologic effects are termed type 2 DCS. Treatment should
involve immediate administration of 100% oxygen, which
facilitates nitrogen washout by the previously explained
principles of Daltons law. Rehydration as well as
advanced cardiac or trauma life-support protocols should
be followed by transfer to a hyperbaric facility emergently. The patient should be treated with hyperbaric
oxygen following U.S. Navy guidelines (Fig. 3), even if
the inspired oxygen and rehydration alone have improved
the patients signs and symptoms because tiny bubbles
may be left that can cause tissue necrosis. Hyperbaric
oxygen shrinks the size of the mostly nitrogen-filled bubbles by the principles of Boyles law, and the increased
pressure also increases the partial pressure of the gas by
Daltons law, hastening complete elimination of the bubble
(Table 2). If the U.S. Navy (Fig. 3) recompression regimen
fails to lead to symptom resolution, the Divers Alert
Network or a medical expert on DCI should be contacted,
and one of a number of recompression tables may be
followed.
PROBLEM WOUNDS
The management of problem wounds should always
include infection control, debridement, aggressive wound
care, and correction of perfusion and oxygenation deficiencies. When an oxygenation deficiency of the wound is
found, in the face of nonreconstructable vascular disease,
hyperbaric oxygen should be considered as an adjunctive
therapy. An increase in tissue oxygen tension by HBO
therapy enhances wound healing by increasing neutrophil bactericidal capacity, inhibiting toxin formation in
and even killing some anaerobes, encouraging fibroblast
activity, and promoting angiogenesis (15). In normal physiology, the oxygen gradient across a wound is essential to
stimulate these components of healing. Oxygen consumption is relatively low in wounds, and microvasculature
damage and peripheral vasoconstriction increase diffusion distances. Partial pressure via Daltons law is the
driving force of diffusion. Hyperbaric oxygen creates a
steep tissue oxygenation gradient, providing a stronger
stimulus than lactate or moderate hypoxia, to initiate and
facilitate wound healing (16,17). These stimulated factors
are thought to include platelet-derived growth factor B
(PDGF-B), Vascular endothelial growth factor (VEGF),
Epidermal growth factor, and others. Several clinical
studies support the use of hyperbaric oxygen to promote
wound healing. Perhaps the studies involving diabetic
lower extremity wounds have been most informative.
Several studies have shown an increased number of
healed wounds, decreased wound size, and decreased
rates of amputation among patients receiving hyperbaric
24
HYPERBARIC MEDICINE
oxygen therapy as an adjunctive treatment (18,19). Baroni et al. reported in a controlled study that a significant
number of subjects receiving HBO went on to heal their
wounds and fewer required amputation when compared
with subjects not receiving HBO (20). In another study
involving 151 diabetic patients with wounds of the lower
extremity, Oriani et al. showed that 130 of these patients
completely healed their wounds with adjunctive HBO
(21). When compared with conventionally treated
wounds, HBO-treated patients had an accelerated rate
of healing, reduced rate of amputation, and an increased
rate of completely healed wounds on a long-term basis
(21). Transcutaneous Oxymetry (TcPO2) is currently the
best tool available to evaluate tissue hypoxia and to select
patients appropriate for HBO therapy. It can also be used
to monitor progress during hyperbaric oxygen therapy.
TcPO2 measurements should be taken with the patient
breathing room air. A value of greater than 50 mmHg
around the wound site indicates that the wound has
adequate oxygenation and hyperbaric oxygen is not likely
to improve healing. Values below 40 at the wound site
should be considered for HBO therapy. Patients with
marginal TcPO2 should be further tested while breathing
100% oxygen. TcPO2 values of greater than 100 while
breathing 100% oxygen is an indicator that they are likely
to respond to HBO therapy. If this challenge TcPO2 is less
than 100, they still may benefit if the tested TcPO2 at the
wound site is greater than 200 mmHg while they are
breathing 100% oxygen at 2.0 ATA in the hyperbaric
chamber (22). A TcPO2 value of less than 30 around a
wound that does not exhibit this response, which indicates
vascular compromise and the patient should be considered for revascularization if possible. Of note, 96% of limbs
with TcPO2 values below 30 mmHg had abnormal arteriograms. It is also important to follow TcPO2 values
weekly, and diabetic patients may have normal or falsely
elevated noninvasive Doppler studies and a low TcPO2,
implying satisfactory perfusion and inadequate oxygenation of the wound and, as such, may pose a diagnostic
delimma. The diabetic patient with normal noninvasive
Doppler and low TcPO2 will respond best to HBO. HBO
therapy should be reserved for those diabetic wounds not
responding to traditional management of debridement,
antibiotics, and general wound care, including vascular
reconstruction. The use of HBO therapy is necessary in
only 1520% of these patients. HBO therapy increases
wound oxygen tension, enhancing host antibacterial
mechanisms and promoting wound healing and is
reserved for wounds in which the primary etiologies are
tissue hypoxia or infection (13). Treatments are delivered
at 2.02.4 atmospheres for 90120 min once or twice daily.
When serious infections are present, patients are typically hospitalized and given IV antibiotics and hyperbaric
treatments twice daily five days a week. The TcPO2 values
should be checked weekly because hyperbaric oxygen
facilitates angiogenesis by a nitric oxide and vascular
endothelial growth factor Beta (VEGF-B). When the room
air TcPO2 is greater than 40 mmHg, the hyperbaric oxygen therapy can safely be discontinued. HBO is an adjuvant treatment; therefore, diabetic control, debridement,
and aggressive wound treatment are given first priority.
When the wound bed has adequate granulation tissue,

application of grafts can shorten morbidity, hospital stay,
and health-care costs. The underlying problem in failure
of a wound to heal is usually hypoxia and infection.
Hyperbaric oxygen treatments in selected patients can
facilitate healing by increasing tissue oxygen tension,
thus providing the wound with a more favorable environment for healing. Therefore, hyperbaric oxygen therapy
can be an important component to any comprehensive
wound care program.
COMPROMISED FLAPS AND GRAFTS
Skin grafts and flaps with adequate blood supply do not
require HBO. Hyperbaric oxygen therapy is extremely
useful in situations where the skin grafts or flaps suffer
from compromised microcirculation or hypoxia.
Flaps
The benefits of HBO on flaps develop from a systemic
elevation in oxygen tension (2325). In addition, HBO
therapy prevents neutrophil adherence and subsequent
vasoconstriction following ischemia. Too often, a compromised flap is allowed to progress over the days following
surgery until visible signs of necrosis obviate the use of
HBO, because delayed treatment with HBO cannot revive
dead tissue. The resulting disappointment, as well as the
associated patient dissatisfaction, can be avoided by rapid
diagnosis of the flap problem and early involvement of the
hyperbaric physician. The keys to successful treatment of
compromised flaps with HBO are accurate diagnosis of the
specific flap problem and appropriate and expedient initiation of hyperbaric oxygen treatment. Awareness of the
different etiologies of flap compromise is necessary to plan
for effective HBO treatment. A random flap with distal
necrosis is completely different from a free flap with total
venous occlusion. Proper classification of flaps, different
etiologies of flap compromise, and understanding of how
HBO is thought to effect ischemia reperfusion injury
defines which patients will benefit from HBO. Flap classification is based on an assessment of blood supply, tissue
composition, and method of movement. Each of these
elements must be evaluated, but it is blood supply that
is most important. The blood supply to the flap is either
axial, based on a named vessel, or random, based on the
subdermal plexus. Commonly, flap compromise occurs
when the surgeon tries to mobilize tissue outside the
defined arterial supply, when there is a pedicle problem
exists, or when free flaps are exposed to prolonged ischemia. The tissue composition of a flap may include skin,
subcutaneous tissue, fascia, muscle, bone, other tissues, or
a combination of these. Flap composition is very important
because different tissue types have different tolerances to
ischemia. For instance, a myocutaneous flap will be more
susceptible to ischemia than a fasciocutaneous flap,
because muscle is much more sensitive to ischemic injury
than fascia and skin (26). In those circumstances where a
prolonged primary ischemia or any secondary ischemia
resulting from vessel thrombosis and revision anastomosis
exists, the flaps will undergo ischemia reperfusion injury.
HYPERBARIC MEDICINE
When treating compromised flaps, a multimodality approach

should be initiated. This approach should include the use
of vasodilators if arterial vasospasm is suspected, removal
of sutures if tension or compression are suspected, dextran
and pentoxifylline for rheological purposes, medicinal and
chemical leeching for venous congestion, and the early
use of hyperbaric oxygen if blood flow can be documented.
The use of HBO therapy is appropriate only when the
flap problem has been defined, documented perfusion of
the flap exists, appropriate surgical salvage measures have
been first considered, and HBO therapy can be performed
in an expedient manner. Specifically with respect to free
flaps, extended primary ischemia time greater than 2 h
or any secondary ischemia time may result in partial or
total flap necrosis. This injury is usually reversible if
recognized early and treated expeditiously. Essentially,
it is ischemia reperfusion injury. Numerous research
studies support the use of HBO in the salvage of compromised free tissue transfers (27,28). A rat free-flap model
showed similar improvement in flap survival (27). A clinical study evaluated free-flap salvage in the face of prolonged primary or any secondary ischemia (28). Salvage
was significantly better in the HBO treatment group vs.
controls, but only if initiated within 24 h. Free flaps compromised by prolonged primary or secondary ischemia
have responded favorably to HBO treatment with complete salvage, in most cases, if HBO is started early. The
treatment regimen is 2.02.4 ATA, 90 min q 8 h x 24 h, then
q 812 h x 48 h (29). Treatment duration is based on clinical
evaluation.
Grafts
Skin grafts are anatomically different from flaps in that skin
grafts lack an inherent blood supply. Skin grafts are composed of avascular tissue that depends entirely on the recipient bed for oxygenation. HBO is useful in preparing the
recipient bed and in promoting healthy granulation tissue to
support split-thickness skin grafts. One controlled study
showed a significant improvement in skin graft survival
from 17% to 64% with the addition of HBO treatment.
Although literature exists to support the use of HBO for
composite grafts, a study by the University of Mississippi
Medical Center found no significant effect of HBO on rat-ear
composite grafts larger than 1 cm (30,31) Further research
is needed to better understand the effects of HBO on
composite graft survival. The rational for use of HBO in
crush injury, compartment syndrome, frostbite, and other
traumatic ischemias is similar to those for compromised
flaps as they are all cases of ischemia and ischemia reperfusion injury.
CRUSH INJURY, COMPARTMENT SYNDROME, AND

OTHER ACUTE TRAUMATIC ISCHEMIAS
These conditions are trauma-related situations in which
the underlying pathophysiology is that of ischemia reperfusion (IR) injury. Ischemia times of greater than 4 h
willresult in some degree of permanent necrosis. The
physiologic basis of IR injury has become better under-
25
stood in recent years. Most of the animal research centers

around the production of oxygen-free radicals. Although
the endothelial xanthine oxidase pathway has received
much attention in the literature (32), more recent evidence supports the fact that neutrophils are a more
important source of oxygen-free radicals via membrane
NADPH oxidase and degranulation. Also, neutrophil
adhesion is felt to cause ischemia reperfusion IR-associated vasoconstriction.
A perceived paradox exists related to HBO for IR injury.
The less-informed observer often does not understand why
HBO improves reperfusion injury and might think HBO
instead increases free radical formation. (An oxygen-free
radical is an oxygen molecules with an unpaired electron
in its outer shell.) During ischemia, ATP is ultimately
degraded to hypoxanthine and xanthine, which are anaerobic metabolites. With reperfusion, oxygenated blood is
reintroduced into the ischemic tissue, and the hypoxanthine and xanthine plus oxygen creates oxygen-free
radicals. Superoxide and hydroxyl oxygen-free radicals
are formed, which can cause extensive tissue damage.
The authors believe that the major mediator of damage
is, in fact, neutrophil adherence to postcapillary venules
significant and progressive vasoconstriction occurs in
arterioles adjacent to leukocyte-damaged venules. Neutrophil adherence and vasoconstriction lead to a low flow state
in the microcirculation and then vessel thrombosis, which
is the endpoint of IR injury. The leukocyte-damaged venule
is thought to be responsible for the arterial vasoactive
response. HBO inhibits neutrophil adherence to the
endothelial cells and thereby inhibits the ultimate thrombosis of microvessels, but the complete mechanism is still
poorly understood, but is thought to involve the elevation
in nitric oxide mediated by an increase in nitric oxide
syntase (33). Free radical formation is not felt to be worsened with HBO as fewer adherent neutrophils actually
exist to contribute to the neutrophil oxygen-free radicalgenerating system.
Treatment with hyperbaric oxygen in the face of IR
injury carried the concern that that providing extra oxygen would increase free radical production and tissue
damage. This query has been resolved by studies that
have shown that HBO actually antagonizes the ill effects
of IR injury in a variety of tissues (3335). One of the first
studies evaluating HBO and IR injury showed that HBO,
immediately upon reperfusion, significantly improved
skin flap survival following 8 h of global ischemia in a
rat axial skin flap model with increased microvascular
blood flow during reperfusion. Free-flap survival
improves with HBO treatment during reperfusion even
following ischemia times of up to 24 h (36). Hyperbaric
oxygen administered during and up to 1 h following 4 h
global ischemia significantly reduced neutrophil endothelial adherence in venules and also blocked the progressive
arteriolar vasoconstriction associated with reperfusion
injury (37). HBO inhibited in vitro beta-2-integrin
(CD18)-induced neutrophil adherence function, but did
not alter other important neutrophil functions such as
oxidative burst or stimulus-induced chemotaxis and
migration. This latter finding is very important, because
HBO, through its action on the CD18 adhesion molecule,
26
HYPERBARIC MEDICINE
blocks the neutrophil adherence associated with IR injury

without interfering with other neutrophil functions that
would increase the risk of infectious complications. Initially, the focus in acute ischemia caused by trauma should
be restoration of blood supply. The authors, therefore,
recommend HBO therapy for all patients with muscle
ischemia time greater than 4 h and skin ischemia time
greater than 8 h. The major effects of IR injury are felt to
occur within the first 47 h of reperfusion. 2 ATA hyperbaric oxygen increases the tissue oxygen tension 1000%.
Treatment protocol is 2.02.5 ATA for 60 min, q 8 h x 24 h,
then q 812 h x 48 h with clinical re-evaluation. If progressive signs of ischemic injury are still present, the
treatment is continued at 2.0 ATA, q 12 h for 23 more
days. Usually, 72 h of treatment is adequate as long as the
first treatment is initiated within 4 h of surgery.
RADIATION TISSUE DAMAGE AND

OSTEORADIONECROSIS
1.2 million cases of invasive cancer are diagnosed yearly,
half of which will receive radiation therapy and 5% of
which will have serious radiation complications, which
represents 30,000 cases per year of serious radiation
sequellae (38). HBO is also well studied for its use in
treating osteoradionecrosis in conjunction with adequate
debridement of necrotic bone. Carl et al. also reported
success is applying HBO to 32 women with radiation
injury following lumpectomy and radiation compared
with controls (39). Feldmeier and his colleagues reviewed
the literature and found no evidence to support the potentiation of malignant cells or the engancement of cancer
growth (40). The treatment protocol is 2.5 ATA for 90 min
daily for 2050 treatments. HBO can also be used as a
radiosensitizer and are as much as three times
more sensitive to radiation kisses than are hypoxic cells
(41).
ACUTE THERMAL BURNS

HBO is approved by the USMS but it is not covered by
Medicare. Gruber demonstrated in 1970 that the area around
and under a third-degree burn was hypoxic and could only be
raised by oxygen at increased pressure (42). HBO has been
found to prevent extension, reduce edema, increase healing
rates, and decrease total cost in several randomized studies
(43,44). HBO is also thought to decrease the rate of burn
sepsis based on several early studies. The controversy, in
part, surrounds current guidelines for early debridement and
grafting of burns. Once excised, a burn no longer exists and
HBO will not be helpful. In case burns are not easily amenable to excision such as flash burns to the face or groin, HBO
may be helpful to prevent extension of the burn and to aid
healing. Treatment must be started within 24 h. The recommended regimen is 2.0 ATA for 90 min every 8 h on the first
day, then every 12 hours for 5 or 6 days.
ACUTE EXCEPTIONAL BLOOD LOSS ANEMIA
Hyperbaric oxygen for treatment of acute blood loss anemia
is reserved for those patients whose anemia is not immediately treatable for practical, disease process, or religious
reasons, which may include warm antibody hemolytic disease, Jehovas Witnesses, those with rare blood types, and
those who refuse transfusion for other personal reasons. As
explained in the physiology section, HBO dramatically
increases the amount of solublized oxygen the blood can
carry. In fact, Boerema showed, in 1955, that pigs could be
exsanguinated to four-tenths of one gram of hemoglobin per
deciliter and be maintained in a hyperbaric environment of
3 ATA without hypoxia. The goal in HBO therapy for these
conditions is to improve the oxygen depth with the daily or
twice daily HBO treatments until the anemia can be
improved. In between the treatments, the patients should
be maintained a lower FIO2 of inspired oxygen if possible to
help reduce oxygen toxicity.
CARBON MONOXIDE POISONING
REFRACTORY OSTEOMYELITIS
Chronic refractory osteomyelitis (CROM) is infection of
the medullary and cortical portions of the bone that
persists or recurs following treatment with debridement
and antibiotics. The principles of treatment are fairly
simple. First, the dead bone is debrided and bone cultures
should be taken along with administration of appropriate
antibiotics. Next, the interface or cicatrix, which separates the compromised bone from adequate blood supply,
is removed. Finally, hypoxia in the wound must be corrected, which may be accomplished by HBO. The treatment protocol is 2.0 ATA for 90 min daily for 2060
treatments. Note that CROM and refractory osteomyelitis
require the longest treatment protocols. HBO is believed
to oxygenate hypoxic/ischemic tissues, augment host antimicrobial responses, augment osteoclastic activity, and
induce osteogenesis in normal and infected bone and
antibiotic synergism.
In 1966, Wada first used HBO to treat survivors of coal mine

disasters with carbon monoxide poisoning and burns. The
modern-day sources of carbon monoxide include automobile
exhaust, home heaters, portable generators, propane
engines, charcoal burners and camp stoves, and methylene
chloride paint strippers. The initial treatment for carbon
monoxide poisoning is 100% oxygen. The administration
of 100% oxygen via a nonrebreather mask facilitates the
dissociation of CO from hemoglobin to approximately 1.5
h. Hyperbaric oxygen delivered at 2.83.0 ATA reduced
the halflife of CO-bound hemoglobin further to 23 min. In
addition, patients who had one hyperbaric treatment for
CO poisoning had 46% neuropsychiatric sequelae at discharge and 50% at one month versus two HBO treatments
at 2.83.0 ATA having 13% at discharge and 18% at one
month. The current recommendation is 3.0 ATA for 90
min with air breaks delivered every 8 h for a total of 3
treatments (called theWeaver protocol). Some authors
still feel one treatment may be adequate (45).
HYPERBARIC MEDICINE
CYANIDE POISONING
Hydrocyanide gas or HCN is formed when any number of
substances burns, including furniture, asphalt, paper, carpeting (nylon), lighting baths (acrylic), plastic(polystyrene),
and insulation (melamine resins). The antidote for cyanide
poisoning begins with breathing 100% oxygen, ATLS protocols, and administration of IV sodium thiosulphate and is
continued with a slow infusion of sodium nitrate and simultaneous HBO therapy if it is available. The sodium nitrate
creates methemoglobin, which can impair the oxygen-carrying capacity of hemoglobin. HBO increases the amount of
oxygen dissolved in plasma and may offer a direct benefit.
The treatment regimin is 3.0 ATA with 30/10 airbreaks.
HYPERBARIC CHAMBER FACILITY DESIGN

AND SAFETY
Over 500 hyperbaric facilities exist in the United States,
and the number of hyperbaric chambers is steadily increasing worldwide. Hyperbaric chambers are classified as
either monoplace or multiplace. They differ functionally
in that the monoplace chamber instills oxygen into the
entire chamber environment, whereas in a multiplace
chamber, patients breathe 100% oxygen via a breathing
mask or oxygen hood and exhaled gases are vented outside
the chamber. Monoplace chambers are constructed either
as an acrylic cylinder with metal ends or are primarily
constructed of metal. Most commonly, the monoplace
chambers are formed from an acrylic cylinder from 20 to
40 inches in diameter with tie rods connecting it to end
caps. The opening is a rotating lock or a cam action lever
closure. Separate oxygen and air sources provide the oxygen sources and air for air breaks during therapy. An
oxygen vent must be exhausted outside the building.
The through ports on the HBO chamber door allow passage
of specially made intravenous monitoring devices and
ventilators. The larger diameter monoplace chambers
are more comfortable; however, they require more oxygen
and can be heavier and more expensive to install. The
acrylic chambers can provide a maximum of 3 ATA pressure. Alternatively, some monoplace chambers are constructed mostly of steel with acrylic view ports, which
can accommodate pressures of up to 6 ATA and are often
used in special situations such as offshore rigs where a
compact chamber is needed to treat decompression illness
required in U.S. Navy Table 5. Multiplace chambers are
much larger and are designed to provide treatment to
multiple people or to manage complex conditions. Some
can even house operating rooms with special precautions.
They are typically made of steel with acrylic view ports
and are designed for operation up to 6 ATA or 165 feet of
sea water. The gauges are reported in feet of sea water on
these multiplace chambers to facilitate the use of dive
tables for staff or patients. These multiplace chambers
are, therefore, best-suited to treat deep water decompression illness. These chambers can accommodate from
2 to 20 people and have variable configurations including
horizontal cylinders, spherical shapes, and rectangular
chambers.
27
The primary professional hyperbaric medicine societies

in the United States are the Undersea and Hyperbaric
Medical Society (UHMS) and the American College of
Hyperbaric Medicine. The UHMS has developed a clinical
hyperbaric medicine facility accreditation program. This
program can be accessed via the UHMS website at http://
www.uhms.org, and it was designed to assure that clinical
facilities are:
1. Staffed with well-trained specialists;
2. Using high quality equipment that is properly
installed, maintained, and operated to the highest
possible safety standards;
3. Providing high quality care;
4. Maintaining proper documentation of informed consent, treatment protocols, physician participation,
training, and so on (46).
Safety Elements for Equipment and Facilities
The American Society of Mechanical Engineers (ASME) and
the Pressure Vessel for Human Occupancy Committee
(PVHO) define the design and fabrication guidelines for
hyperbaric chambers. Although not required in all states
or worldwide, it is accepted as the international standard
(46). Next, the National Fire Protection Association (NFPA)
has established a safety standard for hyperbaric facilities.
The publication, NFPA 99, Safety Standard for Health Care
Facilities, Hyperbaric Facilities, Chapter 20 explains the
details of and criteria for equipment associated with a
hyperbaric chamber facility. The requirements include fire
abatement systems, air quality, and electrical requirements. These requirements apply to any hyperbaric chamber placed within a health-care facility. Each site must have
a safety director. It is important to have only cotton clothing
and to avoid any sources of sparks or static electricity given
the 100% oxygen (Fig. 2). In addition to these guidelines,
hyperbaric chambers are pressure vessels and, as such, are
subject to boiler and pressure vessel laws. They are also
medical devices and, in the United States, are also subject to
FDA rules for class II medical devices. A chamber is required
to have a clearance from the FDA before the device can be
legally marketed or distributed, which is often called a 510 k
clearance, denoting the form on which the clearance must be
submitted. To check on whether a device has received clearance in the United States, one must contact the manufacturer or the Food and Drug Administration (FDA) most
easily via their website, http://www.fda.gov/scripts/cdrh/
cfdocds/cfpmn/dsearch.cfm.
Facilities must develop defined safety protocols and
emergency plans that are available through both the
Undersea and Hyperbaric Medicine Society (UHMS) and
the American College of Hyperbaric Medicine (ACHM).
FRONTIERS AND INVESTIGATIONAL USES
The use of hyperbaric oxygen therapy has, at times, been
surrounded with controversy and spurious claims from
improving athletic performance to slowing the aging
process. It is essential that the hyperbaric medicine
28
HYPERBARIC MEDICINE
physician, staff, and potential patients understand and

follow the principles of evidence-based practice, which
means prescribing HBO therapy for the conditions proven
to benefit from such treatment. The UHMS website, at
www.UHMS. org, and AHCM are good resources for additional information as are numerous publications on hyperbaric medicine such as the hyperbaric medicine textbook
available through the UHMS website. Investigational uses
for hyperbaric oxygen therapy include carbon tetrachloride
poisoning, hydrogen sulfide poisioning, sickle cell crisis,
spinal cord injury, closed head injury, cerebral palsy, purpura fulminans, intraabdominal and intracranial abscess,
mesenteric thrombosis, retinal artery occlusion, cystoid
macular edema, bells palsy, leprosy, lyme disease, stroke
and traumatic brain injury, and brown recluse spider bite.
Some of the many investigational uses for HBO therapy may
have merit, but these must be rigorously studied using welldesigned trials. As the field of hyperbaric medicine continues to advance, so will our understanding of the complex
physiologic effects of delivering oxygen under pressure.
AKNOWLEDGMENT
Special thanks to Bob Bartlet, MD, for his help in preparing this chapter.
BIBLIOGRAPHY
1. Bakker DJ, Cramer FS. Hyperbaric surgery. Perioperative
care. p 2.
2. Behnke AR, Shaw LA. Use of hyperbaric oxygen in treatment
of compressed air illness. Nav Med Bull 1937;35:112.
3. Boerma I. High tension oxygen therapy. Proc Royal Soc Med
1964;57(9):817818.
4. Bakker DJ, Cramer FS. Hyperbaric Surgery Perioperative
Care. p. 67.
5. Jain KK. Physical Physiological, and Biochemical Aspects of
Hyperbaric Oxygenation. Textbook of Hyperbaric Medicine.
Toronto: Hogrefe & Huber Publishers; 1990. p 11.
6. Matos L, Nunez A. Enhancement of healing in selected
problem wounds. In: Kindwall EP, Whelan HT, eds. Hyperbaric Medicine Practice, 2nd ed. Flagstaff AZ; Best; 1999.
7. Hunt TK, Pai MP. The effect of varying ambient oxygen
tensions on wound metabolism and collagen synthesis. Surg
Gynecol Obstet 1972;135:561567.
8. Knighton DR, Halliday BJ, Hunt TK. Oxygen as an antibiotic: A comparison of the effects of inspired oxygen concentration and antibiotic administration on in vivo bacterial
clearance. Arch Surg 1986;121:191195.
9. Zamboni WA, Roth AC, Russell RC, Graham B, Suchy H,
Kucan JO. Morphologic analysis of the microcirculation during
reperfusion of ischemic skeletal muscle and the effect of hyperbaric oxygen. Plast Reconstr Surg 1993;91(6):111023.
10. Zamboni WA, et al. The effect of acute hyperbaric oxygen
therapy on axial pattern skin flap survival when administered during and after total ischemia. J Reconstr Microsurg
1989;5:343347.
11. Muth CM. Gas embolism (Review). New Eng J Med 342:
476482.
12. Divers Alert Network: Report on 1994 Diving Accidents.
Durham, NC: Duke University; 1995.
13. Barnard EEP, Hanson JM, et al. Post decompression shock
due to extravasation of plasma. BMJ 1966;2:154.
14. Powel MR, Spencer MP, Von Ramm OT. Ultrasonic surveillance of decompression. In: Bennett PB, Elliott DH, eds. The
Physiology of Diving and Compressed Air Work, 3rd ed.
London Bailliere: Tindall; 1982. pp 404434.
15. Zamboni WA. Applications of hyperbaric oxygen therapy in
plastic surgery. In: Oriani G, Marroni A, eds. Handbook on
Hyperbaric Medicine, 1st ed. New York: Springer; 1995.
p. 443484.
16. Hunt TK. The physiology of wound healing. Ann Emerg Med
1988;17:12651273.
17. Hunt TK, Hopf HW. Wound healing and wound infection.
Surg Clinics N Am 1997;77(3):587606.
18. Oriani G, Micheal M, Meazza D, et al. Diabetic foot and
hyperbaric oxygen therapy: A ten-year experience. J Hyperbar Med 1992;7:213221.
19. Wattel FE, Mathieu DM, Fossati P, et al. Hyperbaric oxygen
in the treatment of diabetic foot lesions: Search for healing
predictive factors. J Hyperbar Med 1991;6:263267.
20. Baroni G, Porro T, Faglia E, Pizzi G, et al. Hyperbaric oxygen in
diabetic gangrene treatment. Diabetes Care 1987;10:8186.
21. Oriani G, Micheal M, Meazza D, et al. Diabetic foot and
hyperbaric oxygen therapy: A ten-year experience. J Hyperbar Med 1992;7:213221.
22. Strauss MB, Bryant BJ, Hart GB. Transcutaneous oxygen
measurements under hyperbaric oxygen conditions as a predictor for healing of problem wounds. Foot Ankle Int. 2002
Oct; 23(10):9337.
23. Zamboni WA, Roth AC, Russel RC, Nemiroff PM, Casas L,
Smoot EC. Hyperbaric oxygen improves axial skin flap survival when administered during and after total ischemia. J
Reconstr Micro 1989;5:343347.
24. Hunt TK, Pai MP. The effect of varying ambient oxygen
tensions on wound metabolism and collagen synthesis. Surg
Gyn Obstet 1972;135:561567.
25. Niinikoski J, Hunt TK. Oxygen Tension in Human Wounds. J
Surg Res 1972;12:7782.
26. Mathieu D, et al. Pedicle musculocutaneous flap transplantation: prediction of final outcome by transcutaneous oxygen
measurements in hyperbaric oxygen. Plast Reconstr Surg
1993;91:329334.
27. Waterhouse MA, et al. The use of HBO in compromised free
tissue transfer and replantation: A clinical review. Undersea
Hyperb Med 1993;20(Suppl):54 (Abstract).
28. Perrins DJD, Cantab MB. Influence of hyperbaric oxygen on
the survival of split skin grafts. Lancet 1967;1:868871.
29. Persons BL, Zamboni WA. Hyperbaric oxygen in plastic and
reconstructive surgery. In: Bakker DJ, Cramer FS, eds.
Hyperbaric Surgery Perioperative Care. Flagstaff, AZ: Best;
2002.
30. Mazelowski MC, Zamboni WA, Haws MF, Smoot EC, Stephenson LL. Effect of hyperbaric oxygen on composite graft
survival in a rat ear model. Undersea and Hyperbaric Med
Suppl 1995;22:50.
31. McFarlane RM, Wermuth RE. The use of hyperbaric oxygen
to prevent necrosis in experimental pedicle flaps and composite skin grafts. Plast Reconstr Surg 1966; 37:422430.
32. Angel MF, et al. Free radicals: Basic concepts concerning
their chemistry, pathophysiology, and relevance to plastic
surgery. Plast Reconstr Surg 79:990.
33. Lozano DD, Zamboni WA, Stephenson LL. Effect of hyperbaric oxygen and medicinal leeching on survival of axial skin
flaps subjected to total venous occlusion. Undersea Hyperb
Med suppl 1997;24:86.
34. Thom SR, Bhopale V, Fisher D, Manevich Y, Huang PL,
Buerk DG. Stimulation of nitric oxide synthase in cerebral
cortex due to elevated partial pressures of oxygen: an oxidative stress response. J Neurobiol 2002;51(2):85100.
35. Jones S, Wang WZ, Natajaraj C, Khiabani, Stephenson LL,
Zamboni WA. HBO inhibits IR induced Neutrophil CD 18
Polarization by a nitric oxide mechanism. Undersea Hyperb
Med 2002;35 (Suppl):75.
36. Zamboni WA, Roth AC, Russel RC, Nemiroff PM, Casas L,
Smoot EC. Hyperbaric oxygen improves axial skin flap survival when administered during and after total ischemia. J
Reconstr Micro 1989;5:343347.
37. Gimbel M, Hunt TK. Wound healing and hyperbaric oxygen.
In: Kindwall EP, Whelan HT, eds. Hyperbaric Medicine Practice, 2nd ed. Flagstaff, AZ: Best; 1999. p 169204.
38. Bartlett B. Hyperbaric therapy. Radiation Injury 1994:23.
HBO has been shown to increase angiogenesis and blood flow
in previously irradiated tissue or bone. (Marx RE, Ehler WJ,
Taypongsak PT, Pierce LW. Relationship of oxygen dose to
angiogenesis induction in irradiated tissue. Am J Surg 1990;
160:519524.
39. Carl UM, Feldmeier JJ, Schmitt G, Hartmann KA. Hyperbaric
oxygen therapy for late sequelae in women receiving radiation
after breast conserving surgery. Int J Radiat Oncol Biol Phys
2001;49:10291031.
40. Feldmeier JJ. Hyperbaric oxygen: Does it have a cancer
causing or growth enhancing effect. Proceedings of the consensus conference sponsored by the European society for
therapeutic radiology and oncology and the European committee for hyperbaric medicine. Portugal, 2001:129146.
41. Gray KH, Conger AD, Ebert M, Hornsey S, Scott OCA. The
concentration of oxygen dissolved in tissues at the time of
irradiation as a factor in radiotherapy. Br J Radiol 1953;26:
638648.
42. Wada J, Ikeda T, Kamata K, Ebuoka M. Oxygen hyperbaric
treatment for carbon monoxide poisoning and severe burn in
coal mine gas explosion. Igakunoaymi (Japan) 1965;5468.
43. Germonpre P, Reper P, Vanderkelen A. Hyperbaric oxygen
therapy and piracetam decrease the early extension of deep
partial thickness burns. Burns 1996;6:468473.
44. Cianci P, Sato R, Green B. Adjunctive hyperbaric oxygen reduces
length of hospital stay, surgery and the cost of care in severe
burns. Undersea Biomed Research Suppl 1991;18:108.
45. Bartlett R. Carbon monoxide poisoning. In: Haddad M, Shannon M, Winchester J, eds. Poisoning and Drug Overdose, 3rd
ed. New York: WB Saunders Company; 2002.
46. Bakker DJ, Cramer FS. Hyperbaric Surgery Perioperative
Care. Flagstaff, AZ: Best Publishing; 2002.
See also BLOOD GAS MEASUREMENTS; HYPERBARIC OXYGENATION; OXYGEN
MONITORING; PULMONARY PHYSIOLOGY; RESPIRATORY MECHANICS AND GAS
EXCHANGE; VENTILATORY MONITORING.
HARRY T. WHELAN
JEFFREY A. NIEZGODA
MATTHEW C. LEWIS
Medical College of Wisconsin
Milwaukee, Wisconsin
ERIC P. KINDWALL
BERNADETTE CABIGAS
St. Lukes Medical Center
INTRODUCTION
Hyperbaric oxygen (HBO) is simply the delivery of molecular oxygen in very high dosage. Even though experience
29
has shown HBO to be very useful in a number of conditions,

the exact mechanism of action at the molecular level is not
fully understood. Studies done by Thom et al. (1) demonstrated that elevated oxygen tensions stimulated neuronal
nitric oxide synthase (NOS1) and increased steady-state
nitric oxide concentration in their microelectrodeimplanted rodents. Buras et al. (2) in their studies with
human umbilical vein endothelial cells (HUVEC) and
bovine aortic endothelial cells (BAEC) showed that hyperbaric oxygen (HBO) down-regulated intracellular adhesion
molecule 1 (ICAM 1) expression via the induction of
endothelial nitric oxide synthase (NOS3), which proved
beneficial in treating ischemia reperfusion injuries. Other
studies talk about interactions between nitric oxide and
oxygen species and their role in various disease states.
Clearly, interest in HBO is growing.
Boerema (3) introduced hospital use of the hyperbaric
chamber in the late 1950s in Holland, simply to maintain a
semblance of normoxia in patients undergoing cardiac
surgery. Heartlung machines had not yet been invented,
and the use of the chamber made certain kinds of cardiac
surgery possible for the first time. Boerema felt that if
enough oxygen could be driven physically into solution in
the tissues, which he termed drenching, the circulation to
the brain could be interrupted longer than 34 min. It also
rendered surgery on many pediatric patients less risky. For
example, if the normal arterial pO2 in a patient with
Tetralogy of Fallot was 38 mmHg, placing him in the
chamber might raise it to 94 mmHg. Operating on the
patient under hyperbaric conditions posed much less risk
of ventricular fibrillation when the heart or great vessels
were manipulated.
This idea caught on quickly, and soon large surgical
hyperbaric chambers were built in Glasgow, New York, Los
Angeles, Chicago, Minneapolis, and at Boston Childrens
Hospital. By the early 1960s, however, heartlung
machines became more common, and the need to do surgery in the hyperbaric chamber diminished substantially.
Many large surgical chambers were left to gather dust or
were dismantled, as hospital floor space is always at a
premium. During this time the surgeons, who had been
doing most of the research, left the field. Of the nondiving
conditions, only carbon monoxide poisoning and gas gangrene seemed to be likely candidates for hyperbaric oxygen
treatment based on credible research.
In 1969, however, a double-blind controlled study on the
use of hyperbaric oxygen in senility was published in The
New England Journal of Medicine. Results seemed promising, and this initiated the propagation of hyperbaric
quackery. The original investigators made no sweeping
claims for the research, but simply felt that the area
merited further investigation. Eventually, further
research showed that the results of the study reported in
the New England Journal article were a statistical anomaly and could not be reproduced. However, hyperbaric
enthusiasts seized upon the earlier report, and senility
began to be treated in hyperbaric chambers, along with
a host of other diseases. Most of these were not in medical
centers. Fly-by-night clinics suddenly appeared claiming
to cure anything and everything. Patients were treated for
skin wrinkles, loss of sexual vigor, and a host of other
30
maladies. As there were few investigators doing good

research in the area at that time, the field fell into disrepute.
Fortunately, a few legitimate investigators persisted in
their work, looking at the effects of hyperbaric oxygen in
greater detail. Soon it became clear that under hyperbaric
conditions oxygen had some unusual effects. The Undersea
and Hyperbaric Medical Society created a committee to
investigate the field. After careful study, the committee
laid down guidelines for what should be reimbursed by
third-party payers and what conditions should be considered investigational. Their report appeared in 1977 and
was adopted as a source document for Blue Cross/Blue
Shield (4). About the same time, Jefferson C. Davis of the
United States Air Force School of Aerospace Medicine
edited the first textbook in hyperbaric medicine (5). It
was only then that a firm scientific basis was reestablished
for the field, leading to increased acceptance by the medical
community. The number of chambers operating in hospitals has risen dramatically from only 37 in 1977 to > 500
today. The Undersea and Hyperbaric Medical Society
(www.UHMS.org) and the American College of Hyperbaric
Medicine (www.ACHM.org) have taken responsibility for
setting standards in this field and for encouraging additional research. At this time, 13 clinical disorders have
been approved for hyperbaric treatment. They include air
or gas embolism, carbon monoxide poisoning, clostridial
myonecrosis, crush injury or compartment syndrome,
decompression sickness, problem wounds, severe blood loss
anemia, necrotizing soft tissue infections, osteomyelitis,
radiation tissue damage, skin grafts or flaps, thermal
burns and brain abscess.
Remember that hyperbaric oxygen was introduced
initially into hospitals in order to simply maintain normoxia or near-normoxia in patients undergoing surgery.
It was only later, and quite serendipitously that researchers discovered that oxygen under increased atmospheric
pressure gained some of the attributes of a pharmacologic
agent. Oxygen begins to act like a drug when given at
pressures of 2 atm or greater. For example, oxygen under
pressure can terminate lipid peroxidation in vivo (6), it
can enhance the bacteriocidal capabilities of the normal
leukocyte (7,8), and it can stimulate the growth of new
capillaries in chronically ischemic tissue, such as in the
diabetic foot, or in tissue that has undergone heavy radiation. It can reduce intracranial pressure on the order of
50% within seconds of its initiation, and this effect is
additive to that of hypocapnia (911). HBOT can increase
the flexibility of red cells, augmenting the effects of pentoxifylline (12). It can decrease edema formation by a
factor of 50% in postischemic muscle and prevent second-degree burn from advancing to full-thickness injury
(1315). Hyperbaric oxygen has also been shown to hasten
functional recovery of traumatized peripheral nerves by
almost 30% following repair. Many of these discoveries
have been made only in the last decade.
In a number of these areas, we are beginning to understand the basic mechanisms of action, but overall very
little is understood at the molecular level. It is anticipated
that studies involving nitric oxide synthase will provide
insight regarding the elusive molecular mechanistic
explanation. Also, many contributions to our understanding have come from advances made in the biochemistry of
normal wound healing. We understand that normal oxygen pressures are 8090-mmHg arterially, that oxygen
enters our tissues from the capillaries, and that at this
interface carbon dioxide (CO2) is removed. Under hyperbaric conditions, all of this changes. At a chamber pressure of 2.4 atm (ATA), the arterial oxygen pressure (pO2)
reaches 1500 mmHg, immediately saturating the red
blood cells (RBCs). Upon reaching the tissues, these RBCs
never unload their oxygen. At this high partial pressure of
gas, oxygen diffuses into the tissues directly from the
plasma. Returning to the heart, the RBCs are bathed in
plasma with a pO2 of 150200 mmHg. Tissue oxygen
requirements are completely derived from the plasma.
In theory, one might think that this condition could prove
fatal, as red cells no longer can carry CO2 away from the
tissues. However, we are fortunate that CO2 is 50 times
more soluble in plasma than are oxygen and nitrogen, and
the body has a very capable buffering system which
overcomes the loss of the Haldane effect, which is the
increase in CO2 carrying capacity of deoxygenated red
cells (16).
Another factor to be considered is the actual part of the
circulatory system that overcomes the loss of the Haldane
effect. Traditionally, we think of this exchange occurring in
the capillaries. Under very high pressures, however, computer modeling has shown that nitrogen exchange under
pressure (as in deep sea divers) is probably complete by the
time the blood reaches the arteriolar level. Whether this is
true when hyperbaric oxygen is breathed has not yet been
determined. The rate of metabolism under hyperbaric
conditions appears to be unchanged, and the amount of
CO2 produced appears to be about the same as when
breathing air. It would be interesting to know just at what
level oxygen exchange is accomplished in the tissues, as
this might have practical implications when treating people with severe capillary disease.
Oxygen can be toxic under pressure. Pulmonary toxicity and lung damage can be seen at oxygen pressures
> 0.6 atm during chronic exposure. Central nervous system (CNS) toxicity can manifest as generalized seizure
activity when oxygen is breathed at pressures of 3 atm or
greater. The CNS toxicity was first observed by Paul Bert
in 1878, and is termed the Paul Bert Effect (17). Despite
years of research into this phenomenon, the exact underlying or molecular cause of the seizure has not yet been
discovered. There is a generalized vasoconstriction that
occurs when oxygen is breathed at high pressure, reducing blood flow to muscle, heart, and brain by a factor of
20%, as a defense against toxic quantities of oxygen.
The exact mechanism responsible for this phenomenon is
not fully understood.
Central nervous system oxygen toxicity was evaluated
by the Royal Navy. The purpose of this research was to
determine the time until convulsion so that combat swimmers would know their endurance limits under various
conditions. Volunteer research subjects swam in a test
tank using closed-circuit oxygen rigs until convulsion
occurred and thus established safe oxygen tolerance
boundaries.
Also related to the effect of oxygen, the off phenomenon (18) was first described by Donald in 1942. He
observed that seizures sometimes occurred when the chamber pressure was reduced or when a diver surfaced and
oxygen breathing under pressure was suddenly terminated. Lambertsen (19) provided a description of this type
of seizure activity:
The convulsion is usually but not always preceded by the

occurrence of localized muscular twitching, especially
about the eyes, mouth and forehead. Small muscles of
the hands may also be involved, and incoordination of
diaphragm activity in respiration may occur. After they
begin, these phenomena increase in severity over a period
which may vary from a few minutes to nearly an hour, with
essentially clear consciousness being retained. Eventually
an abrupt spread of excitation occurs and the rigid tonic
phase of the convulsion begins. Respiration ceases at this
point and does not begin again until the intermittent
muscular contractions return. The tonic phase lasts for
about 30 seconds and is accompanied by an abrupt loss of
consciousness. It is followed by vigorous clonic contractions
of the muscle groups of the head and neck, trunk and limbs.
As the incoordinated motor activity stops, respiration can
proceed normally.
Within the wound healing community, current doctrine holds that a tissue pO2 of 3040 mmHg is necessary
for adequate wound healing (20,21). Below 30 mmHg,
fibroblasts are unable to replicate or produce collagen.
Additionally, when the pO2 drops < 30 mmHg, leukocytes
are unable to utilize oxidative mechanisms to kill bacteria. We have noted that the tissue pO2 is critical, but
that the actual quantity of oxygen consumed in wound
healing is relatively small. The amount of oxygen used to
heal a wound is only 10% of that required for brain
metabolism.
Production of new collagen is also a requirement for
capillary ingrowth or proliferation (22). As capillaries
advance, stimulated by angiogenic growth factor, they
must be supported by an extracellular collagen matrix to
facilitate ingrowth into tissue. In the absence of new
collagen, capillary ingrowth cannot occur. This effect is
crucial in treating radionecrosis (2325), where the tissue
is primarily hypovascular, and secondarily hypoxic and
hypocellular. It has been discovered that when collagen
production can be facilitated, new capillaries will invade
the previously irradiated area, and healing will then occur.
The tissue pO2 rises to 80% of normal and plateaus;
however, this is sufficient for healing and will even support
bone grafting. Historically, the only means of managing
radionecrosis was to excise the radiated area and bring in
fresh tissue with its own blood supply. New collagen formation and capillary ingrowth also account for the rise in
tissue pO2, which can be achieved in patients with diabetic
foot lesions.
It is now well understood that the stimulus for growth
factor production by the macrophage is hypoxia and/or
the presence of lactic acid (26,27). Wounds managed in
hyperbaric units are typically ischemic and hypoxic.
Periods of relative hypoxia, required for the stimulation
31
of growth factor production, exist between hyperbaric

treatments.
Surprisingly, oxygen levels remain high in tissues for
longer than one would expect following hyperbaric treatment. In a study by George Hart (28) at Long Beach
Memorial Hospital, a mass spectrometer probe was
inserted in the unanesthetized thigh tissues of normal
volunteers. Muscle and subcutaneous tissue pO2 values
in study subjects remained significantly elevated for 23 h
following hyperbaric oxygen treatment. Arterial pO2 was
also measured and found to rise immediately and significantly under hyperbaric conditions but returned to normal
levels within a couple of minutes upon egress from the
chamber (Fig. 1). Thus, multiple daily HBO treatments can
maintain useful oxygen levels for up to 12 h/day.
Mention has been made of enhanced leukocyte killing of
bacteria under hyperbaric conditions. Jon Mader of the
University of Texas-Galveston (29) carried out a rather
simple, but elegant, experiment to demonstrate this. The
fascinating part of this study is that in the evolution of the
human body, a leukocyte has never been exposed to a
partial pressure of 150 mmHg while in tissues. This level
is impossible to attain breathing air. Nevertheless, when
one artificially raises the pO2 far beyond the leukocytes
normal functional parameters, it becomes even more
lethal to bacteria. This is an anomaly, as one rarely can
improve on Mother Nature. Of some interest in this
regard is that if one bites ones tongue, one is never
concerned about possible infection, even though it is a
human bite. Similarly, hemorrhoidectomies rarely, if
ever, become infected. The reason is that the pO2 of the
tissues in and around the oral cavity are very high, and
the pO2 in hemorrhoidal veins is nearly arterial. Tom
Hunt has shown it is impossible to infect tissue that is
injected with raw staphylococci if the pO2 in the same
tissue is > 50 mmHg. Both he and David Knighton have
described oxygen as an antibiotic (30,31).
The reduction of intracranial pressure is facilitated by
vasoconstriction. Experimentally, Rockswold has shown
that mortality can be halved in victims of closed head
injury with Glasgow Coma Scales in the range of 46.
One of the major mechanisms here is a reduction of intracranial pressure while continuing to oxygenate hypoxic
brain (3236). Sukoff et al. (37) administered 100% O2,
1.5 ATA 60 min every 24 h (maximum of 7 h) to severely
brain injured patients. This resulted in a 50% reduction in
mortality.
A paper published by Mathieu (38) has shown that the
flexibility index of red cells can be changed from 23.2 to 11.3
within 15 hyperbaric treatments. This increase in flexibility can prove quite useful in people with narrowed capillaries. However, whether this phenomenon plateaus at 15
treatments, its duration and underlying mechanism are
still unknown.
Nylander et al. (39) demonstrated that following complete occlusion of the blood flow to rat leg for 3 h, postischemic edema could be reduced by 50% if the animals are
promptly treated with hyperbaric oxygen. He also demonstrated that the mechanism for this was preservation of
adenosine triphosphate (ATP) in the cells, which provides
the energy for the cells to maintain their osmolarity. Cianci
32
Oxygenation
Su
cle
(40,41). Yamaguchi, and others have underscored the

importance of ATP in preventing edema in burn. Treatment twice daily has shown to be more efficacious than
treatment once a day.
Zamboni (42) and Gingrass, working at the University
of Southern Illinois, have shown the effects of hyperbaric
oxygen on speeding functional return in peripheral nerve
repair and grafting. At 6 weeks, there is a 28% improvement of function in the affected leg of these rats.
Niezgoda (43) performed a randomized and doubleblinded study in human volunteers investigating the effect
of hyperbaric oxygen in a controlled burn wound model. He
demonstrated statistically significant decreases in edema
formation and wound exudate in the hyperbaric oxygen
treated group.
Finally, the mechanism for the effects of hyperbaric
oxygen in carbon monoxide poisoning (44,45) is now better
understood. Traditionally, it was felt that the mere presence of carboxyhemoglobin blocked transport of oxygen to
the tissues. However, studies by Goldbaum et al. (46) at the
Armed Forces Institute of Pathology in 1975 lead us to
different conclusions. Impairment of cytochrome A3 oxidase and lipid peroxidation occurring following a reperfusion injury are now suggested as the primary pathways in
the pathophysiology causing fatality. Stephen Thom (47
50) at the University of Pennsylvania pioneered research
in this area. It appears that as carbon monoxide levels fall,
the products of lipid peroxidation rise, indicating that
brain damage is occurring only during the resuscitative
phase, thus becoming reperfusion injury. Thom demonstrated that a period of hypotension (even though it may
only be a matter of seconds) is enough to initiate lipid
peroxidation. Oxygen at 1 atm has no effect on halting the
process. However, oxygen at 3 atm terminates lipid peroxidation. Patients who have been treated acutely with
hyperbaric oxygen rarely exhibit signs of delayed deterioration, reported in 3040% of severe cases treated only
with normobaric oxygen. The probable mechanism for this
us
Figure 1. Arterial, muscle and subcutaneous pO2 after HBO treatment.
ne
ou
cu
ta
Plug 1
Pressure (in atmospheres)
B
O
Time (in hours)
is the ability of hyperbaric oxygen at three ATA to terminate lipid peroxidation.

Finally, in many ways it seems paradoxical that oxygen
at high pressure, which intuitively would seem to provide
more substrate for free-radical formation, still benefits
tissues from crush injury and postischemic states. But
ironically, it is precisely this hyperbaric environment that
promotes certain so-called reactive oxygen species with
inherent protective qualities (51). More studies are certainly needed to investigate the underlying pharmacologic
benefits afforded by hyperbaric oxygen. We have only just
begun to explore and utilize a treatment modality whose
time has come.
BIBLIOGRAPHY
1. Thom SR, Fisher D, Zhang J, Bhopale VM, Ohnishi ST, Kotake
Y, Ohnishi T, Buerk DG. Stimulation of perivascular nitric
oxide synthesis by oxygen. Am J Physiol Heart Circ Physiol
2003;284:H12301239.
2. Buras JA, Stahl GL, Svoboda KKH, Reenstra WR. Hyperbaric
oxygen downregulates ICAM-1 expression induced by hypoxia
and hypoglycemia: the role of NOS. Am J Physiol Cell Physiol
2000;278:C292C302.
3. Boerema I, Kroll JA, Meijne NG, Lokin E, Kroon B, Huiskes
JW. High atmospheric pressure as an aid to cardiac surgery.
Arch Chir Neerl 1956;8:193211.
4. Kindwall EP. Report of the committee on hyperbaric oxygenation. Bethesda: Undersea Medical Society; Bethesda 1977.
5. Davis JC, Hunt TK. Hyperbaric oxygen therapy. Undersea
Medical Society; Bethesda 1977.
6. Thom SR. Molecular mechanism for the antagonism of lipid
peroxidation by hyperbaric oxygen. Undersea Biom Res
(Suppl) 1990;17:5354.
7. Andersen V, Hellung-Larsen P, Sorensen SF. Optimal oxygen
tension for human lymphocytes in culture. J Cell Physiol
1968;72:149152.
8. Gadd MA, McClellan DS, Neuman TS, Hansbrough JF. Effect
of hyperbaric oxygen on murine neutrophil and T-lymphocyte
functions. Crit Care Med 1990;18:974979.
9. Hayakawa T, Kanai N, Kuroda R, et al. Response of cerebrospinal fluid to hyperbaric oxygenation. J Neurol Neurosurg
Psych 1971;34:580356.
10. Miller JD, Fitch W, Ledingham IM, et al. Reduction of increased
intracranial pressure. Neurosurgery 1970;33: 287296.
11. Sukoff MH, Ragatz RE. Hyperbaric oxygenation for the
treatment of acute cerebral edema. Neurosurgery 1982;10:
2938.
12. Nemiroff PM. Synergistic effects of pentoxifylline and hyperbaric oxygen on skin flaps. Arch Otolaryngol Head Neck Surg
1988;114:977981.
13. Cianci P, Lueders H, Shapiro R, Sexton J, Green B. Current
status of adjunctive hyperbaric oxygen in the treatment of
thermal wounds. In: Proceedings of the second Swiss symposium on hyperbaric medicine; Baker DJ, JS, editors. Foundation for Hyperbaric Medicine; Basel: 1988. p 163172.
14. Grossman AR. Hyperbaric oxygen in the treatment of burns.
Ann Plast Surg 1978;1:163171.
15. Hart GB, OReilly RR, Broussard ND, Cave RH, Goodman DB,
Yanda RL. Treatment of burns with hyperbaric oxygen. Surg
Gynacol Obstet 1974;139:693696.
16. Coburn RF, Forster RE, Kane PB. Considerations of the
physiological variables that determine the blood carboxyhemoglobin concentrations in man. J Clin Invest 1965;44: 1899
1910.
17. Bert P. Barometric Pressure. 1879. p 579. (translated by
Hitchcock MS, Hitchcock FA) Bethesda: Reprinted by the
Undersea and Hyperbaric Medicine Society; 1978.
18. Donald KW. Oxygen poisoning in man. Br Med J 1947;
712717.
19. Lambertsen CJ. In: Fenn WO, Rahn H, editors. Handbook of
Physiology, Respiration. Washington, D.C.: American Physiological Society; Section 3, Volume II. p 10271046.
20. Knighton DR, Hunt TK, Scheuenstuhl H, Halliday B, Werb Z,
Banda MJ. Oxygen tension regulates the expression of angiogenesis factor by macrophages. Science 1983;221:12831285.
21. Knighton DR, Oredsson S, Banda MJ, Hunt TK. Regulation of
repair: hypoxic control of macrophage mediated angiogenesis.
In: Hunt TK, Heppenstall RB, Pines E, Rovee D, editors. Soft
and hard tissue repair. New York: Praeser; 1948.
p 4149.
22. Knighton DR, Hunt TK, Thakral KK, Goodson WH. Role of
platelets and fibrin in the healing sequence, an in vivo study
of angiogenesis and collagen synthesis. Ann Surg 1982;196:
379388.
23. Davis JC. Soft tissue radiation necrosis: The role of hyperbaric
oxygen. HBO Rev 1987;2(3):153167.
24. Davis JC, et al. Hyperbaric oxygen: A new adjunct in the
management of radiation necrosis. Arch Otol 1979;105:5861.
25. Hart GB, Mainous EG. The treatment of radiation necrosis
with hyperbaric oxygen (OHP). 1976;37:25802585.
26. Jensen JA, Hunt TK, Scheuenstuhl H, Banda MJ. Effect of
lactate, pyruvate, and physician on secretion of angiogenesis
and mitogenesis factors by macrophages. Lab Invest 1986;54:
574578.
27. Knighton DR, Schumerth S, Fiegel VD. Microenvironmental
regulation of macrophage growth factor production. In
preparation.
28. Hart GB, Wells CH, Strauss MB. Human skeletal muscle and
subcutaneous tissue carbon dioxide, nitrogen and oxygen gas
tension measurement under ambient and hyperbaric conditions. J App Res Clin Exper Therap Spring 2003;3(2).
29. Wang J, Corson K, Mader J. Hyperbaric oxygen as adjunctive
therapy in vibrio vulnificus septicemia and cellulites. Undersea Hyperbaric Med Spring 2004, 31(1):179181.
30. Hunt TK, Linsey M, Grislis G, Sonne M, Jawetz E. The effect of
differing ambient oxygen tensions on wound infections. Ann
Surg 1975;181:3539.
33
31. Knighton DR, Halliday B, Hunt TK. Oxygen as an antibiotic. A

comparison of the effects of inspired oxygen concentration and
antibiotic administration on in vivo bacterial clearance. Arch
Surg 1986;121:191195.
32. Miller JD, et al.The effect of hyperbaric oxygen on experimentally increased intracranial pressure. J Neurosurg 1970;32:
5154.
33. Miller JD, Ledingham IM. Reduction of increased intracranial
pressure: Comparison between hyperbaric oxygen and hyperventilation. Arch Neurol 1971;24:210216.
34. Mogami H, et al.Clinical application of hyperbaric oxygenation
in the treatment of acute cerebral damage. J Neurosurg
1969;31:636643.
35. Sukoff MH, et al.The protective effect of hyperbaric oxygenation in experimental cerebral edema. J Neurosurg 1968;29:
236241.
36. Sukoff MH, Ragatz RE. Hyperbaric oxygen for the treatment
of acute cerebral edema. Neurosurgery 1982;10(1):2938.
37. Sukoff MH. Effects of hyperbaric oxygenation [comment]. J
Neurosurg 2001;94(3):403411.
38. Mathieu D, Coget J, Vinkier L, Saulnier F, Durocher A, Wattel
F. Red blood cell deformability and hyperbaric oxygen therapy.
(Abstract) HBO Rev 1985;6:280.
39. Nylander G, Lewis D, Nordstrom H, Larsson J. Reduction of
postischemic edema with hyperbaric oxygen. Plast Reconstr
Surg 1985;76:596601.
40. Cianci P, Lueders HW, Lee H, Shapiro RL, Sexton J,
Williams C, Green B. Adjunctive hyperbaric oxygen reduces
the need for surgery in 40-80% burns. J Hyper Med 1988;3:
97101.
41. Cianci P, Lueders HW, Lee H, Shapiro RL, Sexton J, Williams
C, Green B. Hyperbaric oxygen and burn fluid requirements:
Observations in 16 patients with 40-80% TBSA burns. Undersea Biomed Res (Suppl) 1988;15:14.
42. Zamboni WA, Roth AC, Russell RC, Nemiroff PM, Casa L,
Smoot C. The effect of acute hyperbaric oxygen therapy on
axial pattern skin flap survival when administered during
and after total ischemia. J Reconst Microsurg 1989;5:
343537.
43. Niezgoda JA, Cianci P. The effect of hyperbaric oxygen on a
burn wound model in human volunteers. J Plast Reconstruct
Surg 1997;99:16201625.
44. Brown SD, Piantadosi CA. Reversal of carbon monoxide-cytochrome C oxidase binding by hyperbaric oxygen in vivo. Adv
Exp Biol Med 1989;248:747754.
45. End E, Long CW. Oxygen under pressure in carbon monoxide
poisoning. J Ind Hyg Toxicol 1942;24:302306.
46. Goldblum LR, Ramirez RG, Absalon KB. Joint Committee on
Aviation Pathology XII. What is the mechanism of carbon
monoxide toxicity? Aviat Space Environ Med 1975;46(10):
12891291.
47. Thom SR. Antagonism of carbon monoxide-mediated brain
lipid peroxidation by hyperbaric oxygen. Toxicol Appl Pharmacol 1990;105:340344.
48. Thom SR, Elbuken ME. Oxygen-dependent antagonism of
lipid peroxidation. Free Rad Biol Med 1991;10:413426.
49. Thom SR. Carbon-monoxide mediated brain lipid peroxidation
in the rat. J Appl Physiol 1990;68:9971003.
50. Thom SR. Dehydrogenase conversion to oxidase and lipid
peroxidation in brain after carbon monoxide poisoning. J Appl
Physiol 1992;73:15841589.
51. Thom SR, Bhopale V, Fisher D, Manevich Y, Huang PL, Buerk
DG. Stimulation of nitric oxide synthase in cerebral cortex due
to elevated partial pressures of oxygen: An oxidative stress
response. J Neurobiol 2002;51:85100.
See also HYBERBARIC
MEDICINE; OXYGEN MONITORING.
34
HYPERTENSION.
See BLOOD
PRESSURE MEASUREMENT.
temperature history is known (5),

TDt

R
LEE CHIN
University of Toronto
Toronto, Ontario, Canada
J. CARL KUMARADAS
Ryerson University
Toronto, Ontario, Canada
INTRODUCTION
Interstitial hyperthermia or thermal therapy is a minimally invasive method for the treatment of cancer. Radio
frequency (RF), microwave, laser light, or ultrasound
energy is delivered through one or more thin needle devices
inserted directly into the tumor.
Interstitial devices have the significant advantage
over external devices of being able to deliver thermal
energy directly into the target region, thereby avoiding
depositing energy into intervening nontarget tissue.
Their main disadvantage is that the needle devices
employed often deposit energy over only a small volume.
This can make it challenging to deliver an adequate
thermal dose to large target regions. This problem was
highlighted in an early radiation therapy oncology group
(RTOG) phase III trial in which only 1 out of 86 patients
was deemed to have received an adequate thermal treatment (1).
These early challenges in interstitial hyperthermia
have been addressed, to some extent, through the development of improved heating devices and more detailed
monitoring of applicator placement and dose delivery.
Quality assurance guidelines have been developed by
the RTOG to raise the quality of heating (2). The guidelines recommend pretreatment planning and equipment
checks, the implantation of considerations and documentation, the use of thermometry, and the development of
safety procedures. Treatment procedures have also been
improved through the use of more detailed thermometry,
especially using magnetic resonance imaging approaches
(3,4).
THERMAL DOSE AND HEAT TRANSFER
The goal of interstitial thermal therapy is to deliver a
prescribed dose to a target volume. Thermal dose is defined
as equivalent minutes at 43 8C, or TD. The units of TD are
minutes, which represents the time tissue would need
to be maintained at a constant temperature of 43 8C to
have the same effect as the particular timetemperature
history that the tissue was exposed to. The thermal dose
after & minutes of heating can be calculated if the time
R43Tt dt
where
MICHAEL D. SHERAR
London Health Sciences Centre
and University of Western
Ontario
London, Ontario, Canada
0:25 for T 43 C
0:5 for T > 43 C
1
2
The dose prescribed for treatment depends on whether the

heating is being used as an adjuvant to radiation or systemic therapy, or whether it is being used as a stand-alone
treatment to coagulate tissue. For the former use, the dose
prescribed is typically 1060 min (Eq. 1) and for the latter it
is usually prescribed to be > 240 min (Eq. 2). This is
because temperatures employed for adjuvant treatment
(usually referred to as hyperthermia) are in the 40
45 8C range. For stand-alone coagulation (usually referred
to as thermal therapy or thermal ablation), temperatures
in the range of 5590 8C are used.
The temperature (T) produced in tissue depends on the
heat deposition by the applicator, heat conduction, and
blood flow according to
rc
@T
r
krT v
rT Q
@t
where r is the tissue mass density, r is the heat capacity of

the tissue, k is the thermal conductivity of the tissue, v is
the blood velocity profile, and Q is the heat absorbed per
unit volume. Detailed knowledge of the blood velocity
profile at the capillary level is generally unknown, and
even if it were known the calculations would require
impractically large computational resources. While several
models have been proposed to calculate heat transfer due to
perfusion, the Pennes bioheat transfer equation is most
often employed (6)
rc
@T
r
krT wcb T Tb Q
@t
where w is blood mass perfusion rate, cb is the blood heat

capacity, and Tb is the temperature of the blood entering
the treatment field, and v is the velocity field of any
convective flow (e.g., as the blood in large vessels). This
equation can be used to predict the temperature in tissue,
and therefore plan thermal therapy or hyperthermia treatments if the perfusion rate is known. Pennes equation does
not accurately predict for the effect of large blood vessels
that must be modeled individually.
ELECTROMAGNETIC HEATING
The heat absorbed (or deposited) in tissue is often described
in terms of the power per unit mass. It is called the specific
absorption rate or SAR. For electromagnetic devices heat is
deposited by the motion of charges or ions. The movement
of charge depends on the electric field produced by the
applicator in tissue. In microwave and RF hyperthermia,
the applicators are driven by sinusoidally time-varying
signals. In this case, the electric field can be written in
phasor form E such that the electric field is given by,
Et REe jvt , where R(x) is the real part of the complex
vector x, and v is the angular frequency of the driving
35
signal. The SAR is then

SAR
Q
s
E
E
r 2r
where s is the electrical conductivity of the tissue.

The calculation of the electric field E is based on
Maxwells equations. For microwave devices, these equations are combined to produce the Helmholtz vector wave
equation
2
rrEk E0
where k is the complex-valued wavenumber given by
k2 v2 me jvms and m is the magnetic permeability
of the medium, which for tissue is the same as the
free-space value, and e is the electrical permittivity of
the medium. The divergence free condition, r
E 0,
may have to also be explicitly imposed if the solution
technique does not inherently do this.
For RF devices, the frequency is sufficiently low that the
displacement currents can be ignored. In this case, it is
usually simpler to determine the scalar electric potential V
and from this derive the electric field, E rV. The
electric potential obeys a Poisson-type equation
r
krV 0
For models of both microwave and RF devices, the
governing Helmholtz or Poisson equation is imposed in a
domain with a known electric field or electric potential
specified as a boundary condition to represent the power
source. Another condition that is often imposed on the
surface of metals is that the tangential component of the
electric field is zero, n E 0.
The solution of the governing equations with appropriate boundary conditions is impossible for all but the simplest geometries. For most practical cases, numerical
methods and computational tools are required. The finite
difference time domain (FDTD) method (7), the finite element (FE) method (8,9), and the volume surface integral
equation (VSIE) method (10) are the most commonly utilized methods for solving the governing equations in electromagnetic hyperthermia and thermal therapy. In the
FDTD method, the domain is discretized into rectangular
elements. The accuracy of a FDTD solution depends on the
size of the mesh spacing. Smaller elements produce more
accurate solutions, but also require more memory to store
the system of equations. Since the grids are rectangular,
their nonconformation to curved tissue boundaries produces a stair-casing effect. Therefore, a large number of
elements are required to model such geometries accurately.
Unlike the FDTD method, the FE method uses tetrahedral
meshes in the domain and the VSIE method uses triangular
meshes on domain surfaces. Tetrahedral and triangular
meshes are more suitable than regular finite difference
grids for three-dimensional (3D) modeling since they do
not have the stair casing effect at tissue boundaries.
RADIO FREQUENCY DEVICES
In RF, thermal therapy tissue is heated by electrical resistive (or J) heating. The heating devices, or applicators, are
Figure 1. The heating in RF devices is caused by current flow.

Since the current flows from the heating electrode to the ground
pad, there is a high current density near the electrode due to its
small size compared to the ground pad. This results in heating that
is localized to the heating electrode.
inserted interstitially to produce currents in the tissue. The

currents typically oscillate sinusoidally in the kilohertz or
low megahertz frequency range. As a result, this modality
is often referred to as radio frequency or RF heating. There
devices have an advantage over other interstitial devices in
their simplicity and low cost. They can operate at low
frequency, and therefore do not require complex power
generators. The RF probes, due their simplicity, tend to
have the smallest diameter of all the types of interstitial
heating probes. The RF heating technique has been extensively reviewed by others (1115).
There are several designs of RF interstitial devices,
which may be categorized into three groups. The simplest
design consists of a single electrode at a probe tip (often
referred to as a needle electrode) (9,1620). The current
flows between a single electrode at the end of an applicator
and a large ground plate placed at a distal site. Since the
current flows between a small electrode and a large plate,
the currents are concentrated near the electrodes resulting
in SAR patterns that are localized to the electrodes as
illustrated in Fig. 1.
With these single electrode probes the coagulation diameter is usually limited to 1.6 cm. Therefore several
probes are needed to cover a larger area (21), or a single
probe can be inserted into several locations, sequentially,
during a treatment.
Since it is desirable to avoid the insertion of multiple
interstitial probes, single probes that release multiple
electrodes outward from the probe tip have been designed
to produce large coagulation volumes. Two examples of
these are the Boston Scientific (Watertown, MA; formerly
Radio Therapeutics Corporation, Mountain View, CA) RF
3000 system in which 1012 tines are deployed from a
cannula to form an umbrella shape (Fig. 2) and the RITA
Medical Systems (Mountain View, CA) Starburst probes
with up to 9 tines. In some configurations, some of the tines
in the Starburst probes are replaced with dedicated thermocouples while others are hollow electrodes through
which saline can be infused into the target region to
enhance heating. These multielectrode probes are able to
produce coagulation regions with diameters up to 7 cm,
although complete coverage of a large region can be difficult in high blood flow organs, such as the kidney (22).
The negative RTOG phase III trial, in which only 1 out
of 86 patients was deemed to have received an adequate
thermal dose (1) illustrated the need to not only increase
the target volume coverage, but also to control the heating.
36
Figure 2. A Boston Scientific (Watertown, MA; formerly Radio Therapeutics Corporation,

Mountain View, CA) insterstital RF probe with 10 tines that are deployed from the cannulus
after insertion into a target region. The deployed tines produce a coagulation zone that is larger than
the zone that can be produced by a single electrode probe. The top probe is shown with the tines
undeployed (ready for insertion) and the bottom probe shows the probe with the tines deployed (as
they would be after insertion).
Control is needed to enable the conformation of the heating

to irregularly shaped target volumes while avoiding nearby
organs at risk and to compensate for heterogeneous cooling
by the vasculature (23). Partial control can be achieved by
appropriate positioning of the probes and the adjustment
their power. Further control along the direction of the
probe is also needed (24,25) and multielectrode current
source (MECS) applicators have been developed to provide
this capability (26). The MECS applicators contain several
electrodes placed along their length with the amplitude
and phase of each electrode independently controlled. In
the most common configuration, the electrodes are capacitively coupled (insulated) with the tissue. The electric fields
induced by the electrodes produce currents in the tissue
that cause heating. Since the electrodes are capacitively
coupled, the probes can be inserted into brachytherapy
catheters, for example, making it feasible to add interstitial
heating as a simultaneous adjuvant to brachytherapy
(interstitial radiation therapy). The electric field (and
hence current) may be induced between electrodes on
the same probe or on separate probes, or it may be induced
between the probe electrodes and a grounding plane.
MICROWAVE DEVICES
Microwave applicators can produce larger coagulation
regions than RF applicators due to their radiative nature.
However, the construction of the power generator and
matching circuitry makes these devices more complex,
and therefore more expensive. Due to this, microwave
interstitial hyperthermia has been used less often in the
clinic than RF interstitial hyperthermia.
Ryan et al. reviewed and compared several types of
microwave interstitial applicators (27) and several excellent reviews of microwave interstitial thermal therapy
exist (2832). The two most commonly used devices are
the dipole antenna and the helical antenna. The dipole
antenna is the simplest form of microwave interstitial
antenna (7,8,33). It is usually constructed from a coaxial
cable with the outer conductor removed from an end
section (typically 1 or 2 cm in length) to expose the inner
conductor (Fig. 3). A power generator feeds a sinusoidally
oscillating signal into the cable at one of the ISM frequency bands between 400 MHz and 3 GHz. The inner- and
outer-conductor electrodes at the tip of the coaxial cable

act as an antenna that produces microwaves that radiate
out into the tissue. Tissue is an attenuating medium that
absorbs microwaves, and this absorbed energy is converted
into heat in the tissue.
The radiative or active length of a typical dipole interstitial device is 13 cm. The devices produce a coagulation
region that is ellipsoidal shaped with a large axis of up to
3 cm along the length of the antenna and a small axis of
up to 2 cm diameter. The drawback of the dipole applicator is that the region of highest SAR, or hot spot, is
located at the point at which the outer conductor is cut
away. Therefore, the tips of these antennas have to be
inserted past the center of the target region, and this can
be a problem if the target region is located adjacent to a
critical structure.
A further problem with dipole antennas is that the SAR
patterns are sensitive to the depth to which the antenna is
inserted into tissue (8). A second common microwave
applicator design, referred to as a helical antenna (34
36), has been designed to make the applicator insensitive to
its insertion depth. In this applicator, one electrode is
wrapped in a helix pattern around an exposed coaxial cable
(Fig. 4). The antennas are also designed to extend the
heating pattern along the applicator and toward the tip
of the antenna compared to the dipole antenna. The SAR
pattern from a BSD Medical (Salt Lake City, UT) helical
antenna is shown in (Fig. 5). The antenna was operating at
915 MHz. The measurement was performed using the
thermographic imaging technique (37) and demonstrates
Figure 3. A schematic representation of a microwave interstitial

dipole antenna applicator. The outer conductor of a coaxial cable is
stripped away to produce a radiating section.
37
section area of a coagulation volume has been noted to

increase by a factor of 2.5 in one study (38) and the
coagulation volume diameter was found to increase from
1.2 to 2.4 cm (39). In microwave heating, the cooling is
usually done by passing water or air through the catheter
containing the antenna (29,38,40). In RF heating, cooling
water is passed inside the electrode to cool the tissue near
the electrode (41,42).
In RF heating, it is also possible to increase the coagulation volume by saline injection from the lumen of the
electrode (43). Since saline is electrically conductive, injecting it into the tumor increases the electrical conductivity of
the tumor, and hence the SAR in the tumor. This technique
has not gained popularity due to the inability of control the
flow of saline in the tumor, resulting in irregular and
unpredictable coagulation regions being produced.
Figure 4. Shown here is a BSD Medical (Salt Lake City, UT)
helical microwave applicator. The inner conductor of a coaxial
cable is extended backward in a helical pattern around the
dielectric insulator. There is no connection between the helical
section and the outer conductor.
that the heating extends along the length of the helix and
that the hot spot is close to the tip of the applicator.
Interstitial microwave applicators have the advantage
over RF applicators in the ability to use arrays of applicators to dynamically steer the SAR pattern (33). For large
target volumes, several applicators can be inserted. The
heating pattern can then be adjusted by not only adjusting
the power to each applicator, but also by adjusting the
relative phase of the signal to each applicator. The phase
can be adjusted such that the microwaves produced by
the applicators interfere constructively in regions that
require heating and interfere destructively in regions that
should be spared. The predetermination of the phase
required for each applicator can be calculated during treatment planning. This is a challenging calculation for applications in which tissue is electrically heterogeneous or
the placement of the applicators cannot be accurately
predicted. In these cases real-time monitoring of the treatment is required and a manual or computer run feedback
control is used to set the phase of the applicators to produce
the desired heating profile.
The size of the coagulation volume is limited by the
maximum temperature in the treatment field. Since the
maximum temperature is usually located at the applicator,
it is possible to increase the coagulation volume by cooling
adjacent to the applicator. Using this technique, the cross-
CLINICAL STUDIES WITH MICROWAVE AND RF DEVICES

Interstitial microwave and RF heating systems have been
widely used to clinically treat a variety of tumors in phase I
(34,4447), phase II (18,32,4449) and phase III trials
(1,50). The RF systems have been used to treat a large
range of sites, including brain (45), head and neck (1),
breast (1), myocardium (51), lung (14), liver (11,52), pancreas (18), prostate (48), and kidney (44,53). Microwave
systems have also been used to treat a large range of sites,
including liver (4), prostate (both carcinoma and benign
hyperplasia) (29,36), head and neck (1,32,49), brain
(34,50), breast (1), and other pelvic areas (1). The heat
treatments are used alone (29,36), or combined with external beam radiation (54), interstitial radiotherapy (brachytherapy) (1,32,46), and/or chemotherapy (17). The heat
treatments are used alone (29,36), or combined with external beam radiation (29,36,48,54,55), combined with external
beam radiation (54) or interstitial radiotherapy (brachytherapy) (1,32,46,48,55), and with chemotherapy (17).
The interstitial hyperthermia treatments are usually
administered under ultrasound, CT or MR guidance. During the treatment the hyperechoginicity of microbubbles
that can be produced at sufficiently high temperatures can
provide some real-time ultrasound feedback of the treatment. Posttreatment evaluation can be performed using
contrast enhanced ultrasound, CT or MR. The vasculature
in the coagulated volume is destroyed and the destroyed
volume can be identified as an unenhanced region in the
image (41,56,57).
LASER DEVICES
Figure 5. The normalized SAR pattern along the coronal plane of

a BSD Medical (Salt Lake City, UT) helical applicator operating at
915 MHz. The image was provided courtesy of Claire McCann.
First described in 1983 by Bown (58), Interstitial Laser

Photocoagulation (ILP) [sometimes referred to as Laser
Induced Thermal Therapy (LITT)] involves the use visible
or near infra-red (IR) light delivered through fibre optic
cables to heat tissue for therapeutic purposes. The ILP has
been investigated as an experimental treatment for a
variety of solid tumors including liver, breast, stomach,
pancreas, kidney, lung, and bone (59). The tissue temperature is raised causing coagulation of the target volume.
Similar to the microwave and RF cases, the production of
38
heat in a local volume of tissue results from the amount of

absorbed laser energy, S(r). In biomedical treatments, such
as LITT, it is the total absorbed laser energy that typically
determines the therapeutic outcome. It is equal to the
product of the local fluence rate, f(r) which is the total
photon power over all directions that pass through a point
area of space, and the absorbing characteristics, ma(r) of the
tissue (60):
Sr ma rfr
The absorbed optical energy deposition pattern is governed
by the absorption and scattering characteristics of the
tissue. An absorption event causes the interacting molecule to enter a vibrationalrotational state that results in a
transfer of energy to surrounding molecules that manifests
as a local increase in temperature (61). Absorption occurs
due to interactions with native molecules called chromophores with examples including melanin, hemoglobin, and
water. In a given tissue, the concentration weighted sum of
the absorption of different chromophores leads to its bulk
macroscopic absorption. Scattering refers to a directional
change in light propagation and likely results from differences in the index of refraction in the various cellular
components, such as the between cell membranes and
the extracellular space. Here the scattering is assumed
to be elastic with no change in energy occurring during the
interaction. The statistical quantities that govern light
interactions are the scattering coefficient, ma(cm1) and
absorption coefficient, ma(cm1) and are defined, respectively, as the probability of scattering or absorption per
average distance traveled (also known as the mean free
path). In the case of scattering, consideration is given to the
probability of scatter in a particular direction. An additional parameter known as the anisotropy factor, g, quantifies this directionality by integrating the average cosine
of the scattering probability over all directions. When
g 0, scattering is isotropic. However, in the case of
biological tissues g typically lies within the range of 0.7
and 0.99 meaning that scattering typically occurs in the
forward direction. The reduced scattering coefficient,
m0s ms 1 g, allows light scattering to be approximated
as isotropic although scattering events are actually in the
forward direction. The inverse of the reduced scattering
coefficient is, therefore, the average distance that light
travels before it changes direction from its original direction of propagation (62).
In theory, Maxwells equations could be used to calculate the scattering and absorption of the EM vector fields
due to the underlying tissue components (63). In this case,
the tissue microstructure could be modeled as random
perturbations, e1(r) in the dielectric constant around a
mean value, e0(r), with the total dielectric constant, e(r),
given by the sum of these quantities. However, in practice,
due to the complex and random composition of tissue, a
complete and accurate description of e(r) has yet to be
realized. Instead a more commonly used solution is to
consider light as a stream of neutral particles or photons
with individual quanta of energy that propagate elastically
throughout the medium. This formalism is governed by
radiative transport theory (64), and assumes light to be
monochromatic while ignoring its conventional wave characteristics, such as polarization, diffraction, interference,
and fluorescence. Although incomplete, the photon model
has been shown to be consistent with experimental measurements in turbid media (65).
A commonly employed model of photon propagation is
the Monte Carlo (MC) method (66), which utilizes probability distributions to simulate the propagation of thousands to millions of individual photon packets based on the
optical properties of tissue to arrive at a statistical representation of the overall light distribution. The MC is amenable to heterogeneous and arbitrary geometries and does
not suffer from the limiting assumptions of analytical
solutions. However, its primary disadvantage is the
requirement of long computational times, on the order of
hours to days, to achieve reasonable statistics. Regardless,
with the increasing speed of modern computers, the Monte
Carlo method remains a viable option for photon simulations. The reader is referred to an excellent review by
Roggan and Muller (67) for the implementation of the
MC model for treatment planning of LITT.
Alternatively, one may employ formal solutions to the
governing equations for photon transport. The energy flow
of photons in a scattering and absorbing medium is
described by the radiative transfer equation (RTE) (64).
The RTE is an integro differential equation that describes
the energy conservation of photons within an infinitesimally small volume that result from losses due to absorption and scattering as well as gains arising from photons
scattered from other directions and from the laser source.
Analytical solutions to the RTE are difficult to obtain.
Hence, various approximations have been proposed to
convert the RTE to a more mathematically tractable and
practical form. A standard technique, called the Pn approximation, expands the radiance and source as a finite series
of spherical harmonics to nth order. The P1 approximation
is the simplest of these expansions and in the steady state
is also known as the diffusion approximation (63,64):
*
r2 f r
ma *
1 *
*
r f r S r
D
D
Here f r is the photon fluence rate, while D is the photon

diffusion coefficient given by
D
1
3m0s ma
The primary assumption of the diffusion equation, that is

linear flux anisotropy, is only accurate when the scattering
properties of the medium are much larger than the absorption properties and at locations > 1/m0s from the source. A
number of analytical solutions to the diffusion equation
exist for simple but practical geometries. The solution for a
point source in an infinite homogeneous medium is given
by (63)
*
f r
Po emeff r
4pr
This solution is particularly useful as, assuming an

infinite medium, it may be integrated numerically to provide the light distribution of cylindrical or extended source
of arbitrary geometries. However, it is well known that

tissue optical properties often change from their native
state after undergoing thermal coagulation. This results in
heterogeneities in optical properties that effect the overall
light distribution (68). In such cases, analytical solutions
are available only for the simplest geometries and numerical
methods such as the finite element (69), finite difference
(70), and boundary element method (71) must be employed.
A thorough discussion of these methods was given in the
preceding section for microwaves and their implementation
in the case of photon propagation is the same.
Initially, bare tipped optical fibers were used to deliver
laser light to the tumor. High temperatures immediately
adjacent to the fiber tip cause the tissue to char and form a
zone of carbonization. The charred fiber then acts as a point
heat source and the temperature of the fiber increases
significantly leading to vacuolization of the surrounding
tissue. The volume of coagulation around the fiber grows
until thermal equilibrium is reached at the edges of the
lesion. Here, the conduction of heat from the fiber is
balanced by the tissues ability to remove energy through
blood flow and thermal conduction.
The size of the lesion depends on the thermal conduction
properties of the tissue, but would normally be limited to
2 cm in diameter. Larger tumors require multiple optical
fiber implants to enable complete coverage of the tumor
volume. For example, a 4 cm diameter tumor would
require at least eight fibers to fully coagulate the tumor.
The limitations of the bare tipped fibers have been
addressed in two ways. The first was to employ a line
source geometry instead of a point source. This can be
achieved by using a diffusing tip fiber where light gradually leaks out of the fiber over an extended distance of a few
centimeters. The second approach is to restrict the temperature of the fiber to lower than the charring threshold
by controlling the power delivered to the fiber. If charring is
avoided, light can propagate into the tissue resulting in
heating at a distance from the fiber and a broader SAR
pattern. These two approaches can be combined to achieve
greater lesion volumes from single fibers. Heisterkamp et
al. (72) demonstrated an almost doubling of the coagulated
volume from 4.32 cm3 (bare tipped) to 8.16 cm3 (temperature restricted diffusing tip) using such an approach.
The other major factor that affects the lesion size is the
wavelength of the light used. Somewhat counterintuitively, light that is less absorbed by tissue, results in
greater lesion sizes. This is because the light can penetrate
further into the tissue, and therefore directly heat at
greater distances from the fiber. The availability of high
power sources at two specific wavelengths (810 nm as
produced by diode lasers and 1064 nm as produced by
Nd:YAG lasers) has dominated the development of interstitial laser thermal therapy. Wyman et al. (73) have
shown that 1064 nm light can enable the creation of
greater lesion sizes due to its greater penetration. However, Nd:YAG lasers are large, generally immobile and
inconvenient and so many have adopted 810 nm as the
wavelength of choice due to the availability of compact and
inexpensive sources. More recently 980 nm lasers have
been employed to combine mobility with greater light
penetration (74,75).
39
Differences between Nd:YAG and Diode lasers are only

realized if charring is avoided. Once charring and carbonization has occurred the fiber acts as a point or line heat
source. There is no further light propagation into the tissue
and subsequent heating has no wavelength dependency.
In order to exploit the penetration of light into the tissue,
the fiber tip temperature must be controlled to avoid
charring. Achieving such control is somewhat challenging
as the temperature of the tip can rise very quickly in a positive
feedback loop. As charring begins, the rate of temperature
rise increases that causes an increasing rate of charring.
Robust, automatic feedback control mechanisms are necessary to ensure controlled heating and lesion formation.
INTERSTITIAL ULTRASOUND
The possibility of developing interstitial ultrasound
devices for hyperthermia applications was proposed by
Hynynen in 1992 (76). The initial studies examined various
design parameters including the choice of ultrasound frequency, electric and acoustic power, and catheter cooling.
As Hynynen showed (76), thin interstitial ultrasound
applicators were likely capable of heating perfused tissue
to therapeutic temperatures.
Ultrasound is a high frequency longitudinal pressure
wave that can pass relatively easily through soft tissue.
Consequently, it has been useful as an energy source for
diagnostic imaging where focussed ultrasound radiators
are used to produce high resolution images of soft tissue
abnormalities. During transmission through tissue energy
is lost due to absorption and to a much lesser extent to
scattering. The absorption is caused by friction as the
pressure wave causes relative motion of the tissue components. These frictional forces cause heating that can be
significant if the incident ultrasound power is high enough.
The absorption, a is frequency dependent where
a a fm
and a and m are coefficients that are variable between
tissues although m is 1.5 for most soft tissues. Rapidly
increasing absorption with frequency is the main reason
that the penetration of diagnostic imaging is limited at
very high ultrasound frequencies. Higher penetration is
also the reason that relatively low ultrasound frequencies
are used for ultrasound heating. Typically, frequencies in
the range 0.52 MHz have been used in external focused
ultrasound heating applications. However, this becomes
problematic for interstitial devices that are small and
resonate at high ultrasound frequencies.
Interstitial ultrasound applicators have since been
developed and are usually designed as thin tubular radiators. The radiator consists of a piezoelectric material that
will resonate acoustically at a frequency f determined by
the wall diameter d:
f
v
2d
where v is the speed of sound in the piezoelectric material

(e. g., 4000 m
s1 in the piezoelectric material PZT 4A). For
interstitial applicators, thin radiators are required. A wall
40
thickness of 0.2 mm, for example, would translate into an

operating frequency of 10 MHz (76). The SAR for a
cylindrical applicator is dependent on its dimensions and
the frequency of operation as given by

r 2m f rr0
SAR 2a fI0
e
r0
where a is the ultrasound absorption coefficient in tissue, I0
is the intensity of ultrasound at the applicator surface, r0 is
the radius of the applicator, r is the distance from the
centre of the applicator to the point of interest and m is the
attenuation coefficient of ultrasound that includes absorption and scattering. Skinner et al. (77) have calculated and
compared the SAR of ultrasound, laser, and microwave
applicators assuming a simple cyclindrical radiation pattern for each. The SAR of all these applicators is dominated
by the thin cylindrical geometry so that despite the larger
penetration depth of ultrasound, only slightly larger diameter lesions can be produced. In order to overcome the
limiting geometry, new interstitial ultrasound applicators
have been developed that take advantage of the focusing
ability of ultrasound (78) or that employs acoustic matching that can result in efficient transmission at multiple
frequencies (79).
The development of interstitial ultrasound applicators
is still at the preclinical stage (80,81) although larger,
intracavitary applicators are being applied in the treatment of prostate cancer using a transrectal technique (82).
BIBLIOGRAPHY
1. Emami BC, et al. Phase III study of interstitial thermoradiotherapy compared with interstitial radiotherapy alone in the treatment of recurrent or persistent human tumors: A prospectively
controlled randomized study by the Radiation Therapy Oncology
Group. Inte J Rad Oncol Biol Phy 1996;34(5): 10971104.
2. Emami BP, et al. RTOG Quality Assurance Guidelines for
Interstitial Hyperthermia. Inter J Rad Oncol Biol Phy 1991;
20(5):11171124.
3. Peters RD, et al. Magnetic resonance thermometry for predicting thermal damage: An application of interstitial laser
coagulation in an in vivo canine prostate model. Magn Reson
Med 2000;44(6):873883.
4. Morikawa S, et al. MR-guided microwave thermocoagulation
therapy of liver tumors: Initial clinical experiences using a 0.5 T
open MR system. J Magn Reson Imaging 2002;16(5):576
583.
5. Sapareto SA, Dewey WC. Thermal dose determination in cancer
therapy. Int J Radiat Oncol Biol Phys 1984;10(6): 787800.
6. Pennes HH .Analysis of tissue and arterial blood temperatures
in the resting human forearm. 1948. J Appl Physiol 1998;
85(1):534.
7. Camart JC, et al. New 434 MHz interstitial hyperthermia
system monitored by microwave radiometry: theoretical and
experimental results. Intern J Hypertherm 2000;16(2):95
111.
8. Mechling JA, Strohbehn JW. 3-Dimensional Theoretical SAR
and Temperature Distributions Created in Brain-Tissue by
915 and 2450 MHz Dipole Antenna-Arrays with Varying
Insertion Depths. Intern J Hypertherm 1992;8(4):529542.
9. Uzuka T, et al. Planning of hyperthermic treatment for malignant glioma using computer simulation. Int J Hypertherm
2001;17(2):114122.
10. Wust P, et al. Simulation studies promote technological development of radiofrequency phased array hyperthermia. Int J
Hypertherm 1996;12(4):477494.
11. Haemmerich D, Lee Jr FT. Multiple applicator approaches for
radiofrequency and microwave ablation. Int J Hypertherm
2005;21(2):93106.
12. McGahan JP, Dodd GD, 3rd. Radiofrequency ablation of the
liver: current status. AJR Am J Roentgenol 2001;176(1):316.
13. Friedman MI, et al. Radiofrequency ablation of cancer. Cardiovasc Intervent Radiol 2004;27(5):427434.
14. Lencioni RL, et al. Radiofrequency ablation of lung malignancies:
where do we stand? Cardiovasc Intervent Radiol 2004;27(6):
581590.
15. Gazelle GS, Goldberg SN, Solbiati L, Livraghi T. Tumor
ablation with radio-frequency energy. Radiology 2000;217(3):
633646.
16. Goletti O, et al. Laparoscopic radiofrequency thermal ablation
of hepatocarcinoma: preliminary experience. Surg Laparosc
Endosc Percutan Tech 2000;10(5):284290.
17. Morita K, et al. Combination therapy of rat brain tumours
using localized interstitial hyperthermia and intra-arterial
chemotherapy. Inter J Hypertherm 2003;19(2):204212.
18. Matsui Y, et al. Selective thermocoagulation of unresectable
pancreatic cancers by using radiofrequency capacitive heating. Pancreas 2000;20(1):1420.
19. Aoki H, et al. Therapeutic efficacy of targeting chemotherapy
using local hyperthermia and thermosensitive liposome: evaluation of drug distribution in a rat glioma model. Int J
Hypertherm 2004;20(6):595605.
20. Lencioni R, et al. Radio-frequency thermal ablation of liver
metastases with a cooled-tip electrode needle: results of a pilot
clinical trial. Eur Radiol 1998;8(7):12051211.
21. Haemmerich D, et al. Large-volume radiofrequency ablation of
ex vivo bovine liver with multiple cooled cluster electrodes.
Radiology 2005;234(2):563568.
22. Rendon RA, et al. The uncertainty of radio frequency treatment of renal cell carcinoma: Findings at immediate and
delayed nephrectomy. J Urol 2002;167(4):15871592.
23. Crezee J, Lagendijk JJ. Temperature uniformity during
hyperthermia: the impact of large vessels. Phys Med Biol
1992;37(6):13211337.
24. vanderKoijk JF, et al. Dose uniformity in MECS interstitial
hyperthermia: The impact of longitudinal control in model
anatomies. Phys Med Biol 1996;41(3):429444.
25. VanderKoijk JF, et al. The influence of vasculature on temperature distributions in MECS interstitial hyperthermia:
Importance of longitudinal control. Intern J Hypertherm
1997; 13(4):365385.
26. Kaatee RSJP. Development and evaluation of a 27 MHz multielectrode current-source interstitial hyperthermia system.
Med Phys 2000;27(12):28292829.
27. Ryan TP. Comparison of 6 Microwave Antennas for Hyperthermia Treatment of CancerSAR Results for Single Antennas
and Arrays. Intern J Rad Oncol Biol Phys 1991; 21(2):403413.
28. Roemer RB. Engineering aspects of hyperthermia therapy.
Annu Rev Biomed Eng 1999;1:347376.
29. Sherar MD, Trachtenberg J, Davidson SRH, Gertner MR.
Interstitial microwave thermal therapy and its application
to the treatment of recurrent prostate cancer. Intern J
Hypertherm 2004;20(7):757768.
30. Fabre JJ, et al. 915 MHz Microwave Interstitial Hyperthermia.
1. Theoretical and Experimental Aspects with Temperature
Control by Multifrequency Radiometry. Intern J Hypertherm
1993;9(3):433444.
31. Camart JC, et al. 915 MHz Microwave Interstitial Hyperthermia. 2. Array of Phase-Monitored Antennas. Intern J
Hypertherm 1993;9(3):445454.
32. Prevost B, et al. 915 MHz Microwave Interstitial Hyperthermia. 3. Phase-II Clinical-Results. Intern J Hypertherm 1993;
9(3):455462.
33. Camart JC, et al. Coaxial Antenna-Array for 915 MHz Interstitial HyperthermiaDesign and Modelization Power Deposition and Heating Pattern Phased-Array. IEEE Trans
Microwave Theory Tech 1992;40(12):22432250.
34. Fike JR, Gobbel GT, Satoh T, Stauffer PR. Normal Brain
Response after Interstitial Microwave Hyperthermia. Intern
J Hypertherm 1991;7(5): 795808.
35. McCann C, et al. Feasibility of salvage interstitial microwave
thermal therapy for prostate carcinoma following failed brachytherapy: studies in a tissue equivalent phantom. Phys Med
Biol 2003;48(8):10411052.
36. Sherar MD, et al. Interstitial microwave thermal therapy for
prostate cancer. J Endourol 2003;17(8):617625.
37. Guy A. Analysis of Electromagnetic Fields Induced in
Biological Tissues by Thermographic Studies on Equivalent
Phanton Models. IEEE Trans Biomed Eng 1971;19:205
214.
38. Trembly BS, Douple EB, Hoopes PJ. The Effect of Air Cooling
on the Radial Temperature Distribution of a Single Microwave
Hyperthermia Antenna In vivo. Intern J Hypertherm
1991;7(2):343354.
39. Goldberg SN, et al. Radiofrequency tissue ablation: increased
lesion diameter with a perfusion electrode. Acad Radiol
1996;3(8):636644.
40. Eppert V, Trembly BS, Richter HJ. Air Cooling for an Interstitial Microwave Hyperthermia AntennaTheory and
Experiment. IEEE Trans Biomed Eng 1991;38(5):450460.
41. Goldberg SN, et al. Treatment of intrahepatic malignancy
with radiofrequency ablation: Radiologic-pathologic correlation. Cancer 2000;88(11):24522463.
42. Solbiati L, et al. Hepatic metastases: percutaneous radiofrequency ablation with cooled-tip electrodes. Radiology 1997;
205(2):367373.
43. Livraghi T, et al. Saline-enhanced radio-frequency tissue
ablation in the treatment of liver metastases. Radiology
1997;202(1):205210.
44. Michaels MJ, et al. Incomplete renal tumor destruction using
radio frequency interstitial ablation. J Urol 2002;168(6):2406
2409.
45. Takahashi H, et al. Radiofrequency interstitial hyperthermia
of malignant brain tumors: Development of heating system.
Exper Oncol 2000;22(4):186190.
46. Seegenschmiedt MH, et al. Clinical-Experience with Interstitial
Thermoradiotherapy for Localized Implantable Pelvic Tumors.
Am J Clin Oncol Cancer Clin Trials 1993;16(3): 210222.
47. Seegenschmiedt MH, et al. Multivariate-Analysis of Prognostic Parameters Using Interstitial Thermoradiotherapy
(Iht-Irt)Tumor and Treatment Variables Predict Outcome.
Intern J Rad Oncol Biol Phys 1994;29(5):10491063.
48. van Vulpen M, et al. Radiotherapy and hyperthermia in the
treatment of patients with locally advanced prostate cancer:
Preliminary results. Bju Inter 2004;93(1):3641.
49. Engin K, et al. Thermoradiotherapy with Combined Interstitial and External Hyperthermia in Advanced Tumors in the
Head and Neck with Depth Greater-Than-or-Equal-to 3 Cm.
Intern J Hyperther 1993;9(5):645654.
50. Sneed PK, et al. Thermoradiotherapy of Recurrent Malignant
Brain-Tumors. Int J Radiat Oncol Biology Physics. 1992.
51. Wonnell TL, Stauffer PR, Langberg JJ. Evaluation of Microwave and Radio-Frequency Catheter Ablation in a Myocardium-Equivalent Phantom Model. IEEE Trans Biomed Eng
1992;39(10):10861095.
52. Buscarini L, Buscarini E. Therapy of HCC-radiofrequency
ablation. Hepato-Gastroenterol 2001;48(37):1519.
41
53. Rendon RA, et al. Development of a radiofrequency based

thermal therapy technique in an in vivo porcine model for the
treatment of small renal masses. J Urol 2001;166(1):292298.
54. Blute ML, Larson T. Minimally invasive therapies for benign
prostatic hyperplasia. Urology 2001;58(6A):3340.
55. Van Vulpen M, et al. Three-dimensional controlled interstitial
hyperthermia combined with radiotherapy for locally
advanced prostate carcinomaA feasibility study. Intern J
Rad Oncol Biol Phy 2002;53(1):116126.
56. Belfiore G, et al. CT-guided radiofrequency ablation: a potential complementary therapy for patients with unresectable
primary lung cancera preliminary report of 33 patients.
AJR Am J Roentgenol 2004;183(4):10031011.
57. Cioni D, Lencioni R, Bartolozzi C, Percutaneous ablation of
liver malignancies: Imaging evaluation of treatment response.
Eur J Ultrasound 2001;13(2):7393.
58. Bown SG. Phototherapy in tumors. World J Surg 1983;7(6):
700709.
59. Witt JD, et al. Interstitial laser photocoagulation for the
treatment of osteoid osteoma. J Bone Joint Surg Br 2000;
82(8):11251128.
60. Welch A. The thermal response of laser irradiated tissue.
IEEE J Quantum Electr 1984;20(12):14711481.
61. Boulnois J. Photophysical processes in recent medical laser
developments: A review. Lasers Med Sci 1986;1(1):4766.
62. Wyman D, Patterson M, Wilson B. Similarity relations for the
interaction parameters in radiation transport and their applications. Appl Op 1989;28:52435249.
63. Ishimaru A. Diffusion Approximation, in Wave Propagation
and Scattering in Random Media. New York: Academic Press;
1978. p. 178.
64. Duderstadt JH. Nuclear Reactor Analysis. New York: John
Wiley & Sons; 1976.
65. Rinzema K, Murrer L. Direct experimental verification of light
transport theory in an optical phantom. J Opt Soc Am A
1998;15(8):20782088.
66. Wilson BC, Adam G. A Monte Carlo model for the absorption
and flux distributions of light in tissue. Med Phys 1983;
10(6):824830.
67. Roggan A, Muller G. Dosimetry and computer based irradiation planning for laser-induced interstitial thermotherapy
(LITT). In: Roggan A, Muller G, editors. Laser-Induced Interstitial Thermotherapy. Bellingham, (WA): SPIE Press;
114156.
68. Jaywant S, et al. Temperature dependent changes in the
optical absorptio nand scattering spectra of tissue. SPIE Proc
1882. 1993;
69. Arridge SR, Schweiger M, Hiraoka M, Delpy DT. A finite
element approach for modeling photon transport in tissue.
Med Phys 1993;20(2 Pt. 1):299309.
70. Pogue BW, Patterson MS, Jiang H, Paulsen KD. Initial assessment of a simple system for frequency domain diffuse optical
tomography. Phys Med Biol 1995;40(10):17091729.
71. Ripoll J, Nieto-Vesperinas M. Scattering Integral Equations
for Diffusive Waves. Detection of Objects Buried in Diffusive
Media in the Presence of Rough Interfaces. J Opt Soc of Am A
1999;16:14531465.
72. Heisterkamp J, van Hillegersberg R, Sinofsky E. Heatresistant cylindrical diffuser for interstitial laser coagulation:
Comparison with the bare-tip fiber in a porcine liver model.
Lasers Surg Med 1997;20(3):304309.
73. Wyman DR, Schatz SW, Maguire JA. Comparison of 810 nm
and 1064 nm wavelengths for interstitial laser photocoagulation in rabbit brain. Lasers Surg Med 1997;21(1):5058.
74. McNichols RJ, et al. MR thermometry-based feedback control
of laser interstitial thermal therapy at 980 nm. Lasers Surg
Med 2004;34(1):4855.
42
75. Kangasniemi M, et al. Thermal therapy of canine cerebral

tumors using a 980 nm diode laser with MR temperaturesensitive imaging feedback. Lasers Surg Med 2004;35(1):4150.
76. Hynynen K. The Feasibility of Interstitial Ultrasound
Hyperthermia. Med Phys 1992;19(4):979987.
77. Skinner MG, Iizuka MN, Kolios MC, Sherar MD. A theoretical
comparison of energy sourcesmicrowave, ultrasound and
laserfor interstitial thermal therapy. Phys Med Biol 1998;
43(12):35353547.
78. Chopra R, Bronskill MJ, Foster FS. Feasibility of linear arrays
for interstitial ultrasound thermal therapy. Med Phys 2000;
27(6):12811286.
79. Chopra R, Luginbuhl C, Foster FS, Bronskill MJ. Multifrequency ultrasound transducers for conformal interstitial
thermal therapy. IEEE Trans Ultrason Ferroelectr Freq Control 2003;50(7):881889.
80. Nau WH, et al. MRI-guided interstitial ultrasound thermal
therapy of the prostate: A feasibility study in the canine model.
Med Phys 2005;32(3):733743.
81. Diederich CJ, et al. Catheter-based ultrasound applicators
for selective thermal ablation: Progress towards MRI-guided
applications in prostate. Int J Hyperther 2004;20(7):739756.
82. Uchida T, et al. Transrectal high-intensity focused ultrasound
for treatment of patients with stage T1b-2NOMO localized
prostate cancer: A preliminary report. Urology 2002; 59(3):
394398.
See also BRACHYTHERAPY,
HIGH DOSAGE RATE; HEAT AND COLD, THERA-
PEUTIC; HYPERTHERMIA, SYSTEMIC; HYPERTHERMIA, ULTRASONIC; PROSTATE

SEED IMPLANTS.
R. WANDA ROWE-HORWEGE
University of Texas Medical
School
Houston, Texas
INTRODUCTION
Systemic hyperthermia is deliberate heating of the whole
body to achieve an elevated core temperature for therapeutic purposes. Other terms used are whole-body
hyperthermia, systemic or whole body thermal therapy,
and hyperpyrexia. The goal of systemic hyperthermia is to
reproduce the beneficial effects of fever. Typically, core
body temperatures of 4142 8C are induced for 12 h, or
alternatively 3940 8C for 48 h. Systemic hyperthermia,
by virtue of application to the whole body, aims to alleviate
systemic disease conditions, in contrast to local or regional
hyperthermia that treats only a specific tissue, limb, or
body region.
HISTORICAL BACKGROUND
The use of heat to treat disease goes back to ancient times.
Application of fire to cure a breast tumor is recorded in an
ancient Egyptian papyrus, and the therapeutic value of
elevated body temperature in the form of fever was appreciated by ancient Greek physicians. Hippocrates wrote,
What medicines do not heal, the lance will; what the
lance does not heal, fire will, while Parmenides stated,
Give me a chance to create a fever and I will cure any

disease. In the first century AD, Rufus (also written as
Refus or Ruphos) of Ephesus advocated fever therapy for
a variety of diseases. Hot baths were considered therapeutic in ancient Egypt, Greece, Rome, China, and India
as they still are in many aboriginal cultures today, along
with burying diseased individuals in hot sand or mud.
Hot baths and saunas are an integral part of health
traditions throughout the Orient, in Indian Ayurvedic
medicine, as well as in Eastern European and Scandinavian countries. Following several earlier anecdotal
reports, several nineteenth century German physicians
observed regression or cure of sarcoma in patients who
suffered prolonged, high fevers due to infectious diseases.
This led to efforts to induce infectious fevers in cancer
patients, for example, by applying soiled bandages or the
blood of malaria patients to wounds. The late nineteenth
century New York physician, William Coley, achieved
cancer cures by administration of erysipelas and other
bacterial endotoxins, now known as Coleys toxins, and
attempted to create standardized preparations of these
pyrogens (1). At around the same time, treatment of
syphilis by placing the patient in a stove-heated room,
or a heat box, became commonplace. Successful hyperthermic treatment of other sexually transmitted diseases, such
as gonorrhea, and neurological conditions, such as chorea
minor, dementia paralytica, and multiple sclerosis along
with arthritis, and asthma were widely reported. Interestingly, it was noted by Italian physicians that upon completion of the draining of the Pontine Swamps near Rome by
Mussolini in the 1930s, not only was malaria eradicated,
but the prevalence of cancer in the area was the same as in
the rest of Italy, whereas earlier the whole malariainfected region was noted for its absence of cancer. It
was concluded that the frequent fever attacks common
in malaria stimulated the immune system to prevent the
development of cancers.
The science of hyperthermia became grounded in the
first few decades of the twentieth century when some of the
biological effects of elevated body temperature were elucidated and attempts were made to understand and control
the therapeutic application of heat. Numerous devices
were developed to produce elevated temperatures of the
body, by a variety of physical means. After a shift in focus to
local and regional hyperthermia, there is now a resurgence
of interest in systemic hyperthermia for treatment of cancer,
as well as other systemic diseases. Whole-body hyperthermia treatment is now carried out at several university
centers in the United States, and Europe (Table 1), where
controlled clinical trials are being carried out. Numerous
private clinics, principally in North America, Germany,
Austria, Eastern Europe, Japan, and China also perform
systemic hyperthermia, mostly as part of holistic, alternative, treatment regimens.
PHYSICS OF SYSTEMIC HYPERTHERMIA

As shown schematically in Fig. 1, in order to achieve body
temperature elevation, there must be greater deposition
of heat energy in the body than heat energy lost from
43
Baoding
Changchun
Jiangmen
Shanghai
Shanghai
Shanghai
Taian City
Zhengzhou
Tokyo
Minsk
Berlin
Frankfurt
Munich
Kecskeme t
Bergen
Novosibirsk
Obninsk
Galveston, TX
Houston, TX
Buffalo, NY
Durham, NC
Asia
China
China
China
China
China
China
China
China
Japan
Europe
Belarus
Germany
Germany
Germany
Hungary
Norway
Russia
Russia
North America
United States
United States
United States
United States
Radio frequency RF; infrared IR.

Starting in 2006.
City
Country
University of Texas Medical Branch

University of Texas Medical School
Roswell Park Cancer Institute
Duke Comprehensive Cancer Centerb
Belarus Center for Pediatric Oncology

and Hematology
Ludwig Maximilian University
Charite Medical Center
Krankenhaus Nordwest
Ludwig Maximilian University Hospital Clinic
Institute of Radiology of Kecskeme t
University of Bergen, Haukeland University Hospital
Siberian Scientific Research Institute of Hyperthermia
Medical Radiological Research Center of
Russian Academy
of Medical Sciences, Obninsk
Modern Hospital, Zhengzhou

Luke Hospital
88th Hospital of PLA
Shanghai Jingan Central Hospital
Second Hospital of Baoding

Jilin Tumor Hospital
Guangdong Jiangmen Renmin Hospital
Changhai Hospital, Second Military Medical School
Department of Tumor Hyperthermia Center
Institution
Table 1. Clinical Academic/Regional Systemic Hyperthermia Centers
Joseph Zwischenberger
Joan M. Bull
William G. Kraybill
Zeljko Vujaskovic
Bert Hildebrandt, Hanno

Riess, Peter Wust
Elke Ja ger, Akin Atmata
Harald Sommer
Miklo s Szu cs
Baard-Christian Schem
Roman Tchervov
Yuri Mardynsky
Reimann Ismail-zade
Dingjiu Li
Akira Takeuchi
Yong Peng
Weiping Tao
Chunzhu Yin
Changguo Hong
Wenping Wu
Yajie Wang
Kai-sheng Hou
Principal Investigator
extracorporeal
IR, Heckel
IR, Heckel
IR, Heckel
IR, Aquatherm
IR, Iratherm
IR, OncoTherm
IR, Iratherm
Water bath
HF EM, Yakhta 5
IR, Iratherm
HF EM, Yakhta-5
RF
IR
12 h, 41.842.5 8C
IR, ET-Space
extracorporeal
IR, ET-Space
extracorporeal
2
6
6
6
h,
h,
h,
h,
42.5 8C
40 8C
40 8C
40 8C
1 h, 41.8 8C
1 h, 41.8 8C
1 h, 41.8 8C
1 h, 41.8 8C
43.544.0 8C
12 h, 41.042.3 8C
2 h, 41.842.5 8C 1 h,
42.543 8C
1 h, 41.8 8C
12 h, 41.8 8C 6 h,
39.540 8C
3.56 h, 4040.5 8C
2 h, 41.6 8C 4 h, 42.1 8C
3.56 h, 4040.5 8C
3.56 h, 4040.5 8C
Protocol (time, temp)
RF
RF
IR
Heat Type, Machinea
44
basal metabolic rate and core temperature has been

determined as
Basal MR
85 1:07Tcore
0:5
which can be exploited to maintain elevated body temperatures

(2). At room temperature a human body produces 84 W,
which increases to 162 W at a core temperature of 41.8 8C.
Heat losses from the body are often termed sensible
(convective, conductive, radiative) and insensible (evaporative, latent). The primary mode of heat loss from the body is
by radiation, as described by the StefanBoltzmann law,
Q0rad
eskin sAskin Tskin Ts 4
DV
Figure 1. Schematic of heat balance mechanisms in the human

body. Body temperature is determined by the balance of metabolic
heat production plus heating from external sources, and heat
losses by radiation, evaporation, convection, and conduction.
conduction, convection, radiation and evaporation, that is,

Q0dep Dt > Q0loss Dt
where Q0 DQ/Dt represents the change in heat energy,

Q (measured in Joules or calories), over a time period
Dt. Net heat energy deposition in a volume element DV
of tissue of density rtis results in an increase in temperature DT dependent on the specific heat of the tissue,
ctis,
!
Q0dep Q0loss

Dt rtis DVctis DT
DV
DV
DT Q0dep Q0loss
Dt
rtis ctis
Heat deposition is the sum of the absorbed power density, Pabs, from external sources and heat generated by
metabolism, Qmet,
DQ0dep
DV
Pabs x; y; z; t
DQ0met
x; y; z; t
DV
If the air temperature is higher than the body surface

temperature, heat is absorbed from air surrounding the
body by the skin, as well as during respiration. Power
deposition in tissue from external electromagnetic fields
depends on the coupling of the radiation field (microwave, RF, ultrasound, visible or IR light) with tissue.
The bodys metabolic rate depends on the amount of
muscular activity, the temperature, pressure and
humidity of the environment, and the size of the body.
Metabolic rate increases nonlinearly with core body
temperature, in part due to the exponential increase
of the rate of chemical reactions with temperature
(Arrhenius equation). An empirical relationship between
where Q0 rad/DV is the power radiated, eskin is the emissivity

of the skin (radiating material), s is Stefans constant 5.6703 108 W
m2/K, Askin is the skin surface
area, Tskin is the temperature of the skin (radiator), and Ts
is the temperature of the surroundings (e.g., air, water,
wax). Human skin is a near perfect radiator in the IR, with
an emissivity of 0.97. At room temperature, >50% of the
heat generated by metabolism is lost by radiation; a clothed
adult loses some 50 W at room temperature. This increases
to 66% at a core temperature of 41.8 8C, as is targeted in a
number of systemic hyperthermia protocols, when the skin
temperature rises to 3940 8C (3).
Direct transfer of body heat to the molecules around the
body (typically air) occurs by conduction, or molecular
agitation within a material without any motion of the
material as a whole, which is described by Fouriers law,
DQ0cond
DT
kAskin
Dx
DV
where DQcond is the heat energy transferred per unit

volume in time Dt, k is the thermal conductivity (W
mK1)
of the material surrounding the body (air, water), and DT is
the temperature difference across thickness Dx of the
material. Air is a poor thermal conductor, therefore heat
loss by conduction is relatively low. On the other hand,
water has a thermal conductivity 20 times that of air at
0 8C, increasing further with temperature, therefore during hyperthermia it is important that any water in contact
with the skin is not at a lower temperature. The relative
thermal conductivity of body tissues is important in determining thermal conduction within the body from external
sources of heat. For example, fat is a relative thermal
insulator with a thermal conductivity one third of that of
most other tissues, therefore fat bodies are slower to heat.
Convective heat transfer involves material movement
and occurs principally via blood moving heat to, or from,
the skin and other tissues, and air currents (respiratory
and environmental) moving warm air to or from the body.
Equation 7 is written for the blood,
DQ0conv
rb cb wb x; y; z; T
T Tb Ub x; y; z; T
rT
DV
7
where wb is the specific capillary blood flow rate, Ub is the
specific blood flow through other vessels. In the context of
systemic hyperthermia, where a patient is in a closed

chamber, environmental air currents can be minimized.
Heat loss by respiration, however, can amount to almost
10% of metabolic heat generation.
Another route of heat loss from the body is evaporation
of perspiration from the skin. Because of the very large
heat of vaporization of water, cooling of the blood in skin
capillaries occurs due to evaporation of sweat. Evaporation
from exhaled moisture also results in cooling of the surrounding air.
DQ0evap
Lv
mw
DV
Dt
where mw is the mass of the water and Lv is the latent heat

of vaporization (2.4 106 J
kg1 at 34 8C). In hot conditions with maximal rates of evaporation, heat loss through
evaporation of sweat can be as much as 1100 W. Heat loss
in the lungs is 10 W.
Combining the heat generation and heat loss terms
leads to a general heat transfer equation, an extension
of the classic Pennes bioheat transfer equation.
0

DQmet
Pabs
DV
0

DQrad DQ0cond DQ0conv DQ0resp DQ0evap

DV
DV
DV
DV
DV
rtis ctis DT
9()
into which the expressions given in Eqs. 28 may be substituted. Precise solution of this equation for temperature
distribution is complex and requires a number of simplifying
assumptions which have generated significant controversy
in bioheat transfer circles. Modeling of temperature distributions within a body subjected to hyperthermia is also
complex because of the heterogeneity of thermal characteristics between and within tissue, the directionality of power
application, and the dynamic nature of thermoregulation by
human body. Nonetheless, the factors governing systemic
heating of the body can be appreciated.
INDUCTION OF SYSTEMIC HYPERTHERMIA
Apart from the induction of biological fever by pathogens or
toxins, all methods of hyperthermia involve transfer of
heat into the body from an external energy source. The
required net power to raise the temperature of a 70 kg
human from 37 to 41.8 8C (2) is 400 W (5.7 mW). While the
heat absorption from these sources is highly nonuniform,
distribution of thermal energy by the vascular system
quickly results in a uniform distribution of temperature.
Indeed, systemic hyperthermia is the only way to achieve
uniform heating of tissues. Because physiological thermoregulation mechanisms such as vasodilation and perspiration counteract attempts to increase core body temperature,
careful attention must be paid to optimizing the physical
conditions for heating such that there is efficient deposition
of heat energy in the body and, even more importantly,
minimization of heat losses. Wrapping the body in reflective
blankets, foil, or plastic film to reduce radiative and evaporative losses, or keeping the surrounding air moist to
45
minimize losses by perspiration are key techniques for

achieving a sustained increase in body temperature.
Noninvasive methods of heating include immersing the
body in hot water or wax, wrapping the body in a blanket or
suit through which heated water is pumped, placing the
patient on a heated water mattress, surrounding the body
with hot air, irradiating with IR energy, and applying RF
or microwave electromagnetic energy. These techniques
may be applied singly or in combination. For example, the
PompSiemens cabinet used until recently throughout
Europe, as well as in the United States, a modification
of a device originally developed by Siemens in the 1930s,
has the patient lying on a heated water mattress under
which an inductive loop generates an RF field, all inside a
chamber through which hot air is circulated. The Russian
Yakhta-5 system applies a high frequency (13.56 MHz)
electromagnetic field through a water-filled mattress to
permit whole body heating up to 43.5 8C and simultaneous
deep local hyperthermia through additional applicators providing 40.6 MHz electromagnetic radiation. The majority of
whole-body hyperthermia systems currently in clinical use
employ IR radiation to achieve systemic heating. Invasive
approaches to systemic hyperthermia are extracorporeal
heating of blood, removed from the body via an arteriovenous
shunt, prior to returning it to the circulation, as well as
peritoneal irrigation with heated fluid (4). A useful schematic
summary of whole-body hyperthermia induction techniques
along with references is provided by van der Zee (5).
All of these approaches involve a period of steady
temperature increase, followed by a plateau or equilibrium
phase where the target temperature is maintained for anywhere from 30 min to several hours, and finally a cool-down
phase. Depending on the method of hyperthermia induction,
the patient may be anesthetized, consciously sedated, administered analgesia, or not given any kind of medication at
all. An epidural block is sometimes given to induce or
increase vasodilation. During radiant heat induction, the
temperature of the skin and superficial tissues (including
tumors) is higher than the core (rectal) temperature
whereas during the plateau (maintenance) phase, the
skinsuperficial tissue temperature drops below the core
temperature. As already described, heat losses due to physiological mechanisms limit the rate of heating that can be
achieved. When insulation of the patient with plastic foil
was added to hot air heating, the heating time to 41.8 8C was
decreased from 230 to 150 min (65%), and further to 110 min
(48%) by addition of a warm water perfused mattress (5).
The homogeneity of the temperature distribution was also
significantly increased by the addition of insulation and the
water mattress. Noninvasive systemic hyperthermia methodologies typically produce heating rates of 110 8C
h1
with 23 8C
h1 being most common. More rapid heating
can be achieved by the invasive techniques, at the expense of
greater risk of infection and morbidity.
COMMERCIALLY AVAILABLE WHOLE-BODY

HYPERTHERMIA SYSTEMS
A number of commercially available devices have resulted
from the development of these initially experimental
46
systems. The SiemensPomp system has already been mentioned, but is no longer commercially available. Similarly,
neither the radiant heat chamber developed by Robins (3),
and marketed as the Aquatherm system, nor the similar
Enthermics Medical Systems RHS-7500 radiant heat
device, both producing far IR radiation (IR C) in a moist
air chamber, are currently being sold, though they are
still in use in several centers. A close relative is the
Iratherm2000 radiant heat chamber originally developed
by von Ardenne and co-workers (6). In this device, waterfiltered infrared radiators at 2400 8C emit their energy from
above and below the patient bed, producing near-IR (IR A)
radiation that penetrates deeper into tissue than far IR
radiation, causing direct heating of the subcutaneous capillary bed. Thermal isolation is ensured by reflective foils
placed around the patient. However, note that significant
evaporative heat loss through perspiration can be a problem
with this system. Also with a significant market share is the
Heckel HT 2000 radiant heat device in which patients lie on
a bed enclosed within a soft-sided rectangular tent whose
inner walls are coated with reflective aluminum foil that
ensures that the short wavelength infrared A and B radiation emitted by four radiators within the chamber uniformly
bathes the body surface. Once the target temperature is
reached, the chamber walls are collapsed to wrap around the
body, thereby preventing radiative and evaporative heat
loss, and permitting maintenance of the elevated temperature, as shown in Fig. 2.
Another radiant heat device, used mainly in Germany,
is the HOT-OncoTherm WBH-2000 whole-body hyperthermia unit which is a chamber that encloses all but the
patients head. Special light-emitting diode (LED) radiators deliver computer-generated, alloy-filtered IR A wavelengths that penetrate the skin to deliver heat to the
capillary bed. The manufacturer claims that these wavelengths also preferentially stimulate the immune system.
Recently, Energy Technology, Inc. of China has released
the ET-SPACE whole-body hyperthermia system, which
Figure 2. Heckel HT-2000 radiant heat whole body hyperthermia

system. Unit at the University of Texas Medical School at Houston.
Patient is in the heat maintenance phase of treatment, wrapped in
the thermal blankets which form the sides of the chamber during
active heating.
produces IR A radiation in a small patient chamber into

which warm liquid is infused to help increase the air
humidity and thereby reduce perspiration losses. A number of low cost, far infrared, or dry, saunas are being sold
to private clinics, health clubs, and even individuals for
treatment of arthritis, fibromyalgia, detoxification, and
weight loss. Examples are the Smarty Hyperthermic
Chamber, the TheraSauna, the Physiotherm, and the
Biotherm Sauna Dome. Table 2 summarizes features of
these commercially available whole-body hyperthermia
devices.
BIOLOGICAL EFFECTS OF SYSTEMIC HYPERTHERMIA
An understanding of the biological effects of systemic
hyperthermia is critical to both its successful induction
and to its therapeutic efficacy. Systemic responses to body
heating, if not counteracted, undermine efforts to raise
body temperature, while cellular effects underlie both
the rationale for the use of hyperthermia to treat specific
diseases, and the toxicities resulting from treatment.
Although improved technology has allowed easier and
more effective induction of systemic hyperthermia, most
of the recent clinical advances are due to better understanding and exploitation of specific biological phenomena.
Physiological Effects of Elevated Body Temperature
The sympathetic nervous system attempts to keep all parts
of the body at a constant temperature, tightly controlled by
a central temperature set point in the preopticanterior
hypothalamus and a variety of feedback mechanisms. The
thermostat has a circadian rhythm and is occasionally
reset, for example, during fever induced by infectious
agents and endotoxins, but not in endogenously induced
hyperthermia. Occasionally, it breaks down completely as
in malignant hyperthermia or some neurological disorders
affecting the hypothalamus. Ordinarily, when core body
temperature rises, the blood vessels initially dilate, heart
rate rises, and blood flow increases in an effort to transport
heat to the body surface where it is lost by radiation,
conduction, and convection. Heart rate increases on average
by 11.7 beats
min1
8C1 and typically remains elevated for
several hours after normal body temperature is regained.
Systolic blood pressure increases to drive the blood flow, but
diastolic pressure decreases due to the decreased resistance
of dilated vessels, thus there is an increase in cardiac
output. Heart rate and blood pressure must therefore be
monitored during systemic hyperthermia, and whole-body
hyperthermia is contraindicated in most patients with
cardiac conditions. Interestingly, hyperthermia increases
cardiac tolerance to ischemia/reperfusion injury probably
due to activation of manganese superoxide dismutase
(Mn-SOD) and involvement of cytokines.
Respiration rate also increases and breathing becomes
shallower. Perspiration results in evaporation of sweat
from the skin and consequent cooling, while the respiration
rate increases in order to increase cooling by evaporation of
moisture from expired air. Weight loss occurs despite fluid
intake. There is a decrease in urinary output and the urine
has a high specific gravity, concentrating urates and
47
http://www.eti.com.cn/EN/pro/product2.htm
http://www.heckel-medizintechnik.de/
frameset_e.html
http://www.hot-oncotherm.com/
oncothermia.htm
http://www.ardenne.de/med_eng/
Energy Technology
Heckel Medizintechnik GmbH
Von Ardenne Institut fu r

Angewandte
Medizinische Forschung, GmbH
Hot-Oncotherm
Website
Manufacturer
6 IR radiators (IR A)
Iratherm 1000
Iratherm 2000
WBH-2000
HT 2000 M
Iratherm 800
Heating Mechanism
Multiple IR radiators
(IR A)
4 300W IR radiators
(IR A, B)
Multiple LED
radiators (IR A)
ET-SPACE
Device Name
Table 2. Commercially Available Clinical Whole-Body Hyperthermia Devices
3739
3742
3738
3742
38.540.5
3941.8
Temperature Range, 8C
Physical medicine,
complementary
medicine, oncology
Oncology,
rheumatology
Oncology
Oncology
Application
48
phosphates. In endogenously induced hyperthermia, but

not in fever, glomerular filtration, as evidenced by the
creatinine clearance, decreases with increasing temperature. As already mentioned, metabolic rate increases nonlinearly with temperature, which leads to an increase in
blood sugar, decreased serum potassium levels, and
increased lactic acid production. All the above normal
physiological effects may be enhanced or counteracted by
anesthesia or sedation, as well as by disease states such as
cancer because of drugs used in treatment or intrinsic
pathophysiological consequences of the disease.
At 42.5 8C, the normal thermocompensatory mechanisms break down and the body displays the symptoms of
advanced heat stroke, namely, lack of sweating, rapid
heart beat, CheyneStokes breathing, central nervous
system disfunction, and loss of consciousness. Ultimately,
breathing ceases despite the continuation of a heart beat.
A variety of effects in the cell nucleus also occur at high

temperatures (>41 8C) including damage to the nuclear
membrane, increases in nuclear protein content, changes
in the structure of nucleoli, inhibition of DNA synthesis
and chromosomal damage in S-phase. These changes in
nuclear structure compromise nuclear function and may
cause cell death, though they are unlikely to be significant
at the temperatures achieved in systemic hyperthermia.
Disaggregation of the spindle apparatus of mitotic cells
may be responsible for the high thermal sensitivity of cells
in mitosis, as well as in S phase. Hyperthermic inactivation
of polymerase b, an enzyme primarily involved in DNA
repair, is sensitized by anesthetics and may have a role to
play in the enhancement of the effects of ionizing radiation
by systemic hyperthermia, as well as in augmenting the
cytotoxic effect of drugs that cause DNA damage.
Metabolic Effects
Cellular Thermal Damage

When temperature is increased by a few degrees Celcious,
there is increased efficiency of enzyme reactions (Arrhenius equation), leading to increased metabolic rates, but at
temperatures > 40 8C molecular conformation changes
occur that lead to destabilization of macromolecules and
multimolecular structures, for example, to the side chains
of amino acids in proteins, which in turn inhibit enzyme
action. Small heat shock proteins (HSP) interact with the
unfolding proteins to stabilize them and prevent their
aggregation and precipitation. Eventually, however, at
42 8C, complete denaturation of proteins begins that
totally disrupts many molecular processes, including deoxyribonucleic acid (DNA) repair. Thus systemic hyperthermia can have significant effects when paired with drugs
that cause DNA damage (e.g., for chemotherapy of cancer).
Membranes are known to be extremely sensitive to
heat stress because of their complex molecular composition of lipids and proteins. At a certain temperature, lipids
change from the tightly packed gel phase to the less
tightly packed liquid crystalline phase, and permeability
of the cell membrane (membrane fluidity) increases. As
temperature increases further, the conformation of proteins
also becomes affected, eventually resulting in disorderly
rearrangement of the lipid bilayer structure and receptor
inactivation or loss. Temperature changes of 5 8C are
necessary to cause measurable changes in normal cell
membrane permeability. Heat-induced cell membrane permeability can be exploited to increase drug delivery, for
example, transdermally, or into tumor cells. Increased vascular permeability due to thermal increase of endothelial
gap size also aids drug delivery into tumors. At higher
temperatures, heat damage to membranes can cause cell
death, but it will also interfere with therapeutic approaches
that depend on membrane integrity (e.g., receptor targeted
drug delivery, antibodies, etc.). Irreversible disruption of
cytoplasmic microtubule organization and eventual disaggregation, as well as disruption of actin stress fibers and
vimentin filaments, occur at high temperatures (4345 8C)
above those used in whole-body hyperthermia, but these
cytoskeletal effects are of concern with loco-regional hyperthermia.
Moderate increases in temperature lead to increased cellular reaction rates, which may be seen as increased oxygen consumption and glucose turnover. In consequence,
cells may become deprived of nutrients, the intracellular
ATP concentration falls, accumulation of acid metabolites
increases pH, and thermal sensitivity increases. Such
conditions are found in tumors and may contribute to their
sensitivity to heat. Further acidifying tumor cells during
hyperthermic treatment seems a promising approach as is
discussed further below. At high temperatures, the citric
acid cycle may be damaged leading to other acidic metabolites. Increased plasma acetate has been measured following clinical whole-body hyperthermia treatments,
which reduces both release of fatty acids from adipose
tissue into plasma and subsequent lipid oxidation.
Endocrine Function
Increases in plasma levels of an array of hormones have
been noted after whole-body hyperthermia. Increased
ACTH levels appear to be accompanied by increased levels
of circulating endorphins. This may explain the sense of
well-being felt by many patients after systemic hyperthermia treatment, and the palliative effect of hyperthermia
treatments for cancer. Increased secretion of somatotropic
hormone after systemic hyperthermia has also been measured (7).
Thermal Tolerance
Thermal tolerance is a temporary state of thermal resistance, common to virtually all mammalian cells, which
develops after a prolonged exposure to moderate temperatures (4042 8C), or a brief heat shock followed by incubation at 37 8C, and also certain chemicals. The decay of
thermotolerance occurs exponentially and depends on the
treatment time, the temperature, and the proliferative
status of the cells. Several days are usually required for
baseline levels of heat sensitivity to be regained, which has
important implications for fractionated therapy. When pH
is lowered, less thermal tolerance develops, and its decay is
slower. Thus the long periods at moderate temperature
achieved by clinical systemic hyperthermia systems should
Effect of Hyperthermia on Tumors
2.0
41.5C/37C uptake ratio
49
1.0
0.0
100 nm
210 nm
Liposome size
Figure 3. Increase in tumor uptake of large liposomes after 1 h

of 41.5 8C whole-body hyperthermia. Systemic heat treatment
increased the effective pore size from 210 to 240 nm. Because
of the large pore size in MTLn3 tumors, 100 nm (average diameter)
liposomes were able to pass into the tumor equally well at normal
and elevated temperatures. The increased effective pore size due
to hyperthermia allowed larger 200 nm liposomes, which were
partially blocked at normal temperatures, to pass more effectively
into the tumor.
induce thermal resistance in normal cells, while the acidic

parts of tumors should be relatively unaffected. This has
not, however, been studied clinically. The mechanisms
involved in the induction of thermotolerance are not well
understood, but there is mounting evidence that heat shock
proteins are involved.
Step-Down Sensitization
Another distinct phenomenon is step-down sensitization in
which an exposure of cells to temperatures >43 8C results
in increased sensitivity to subsequent temperatures of
42 8C or lower. This can be important clinically for local
and regional hyperthermia if there are marked variations
in temperature during the course of treatment, the magnitude of the effect depending on the magnitude of the
temperature change. It has been suggested that this phenomenon could be exploited clinically by administering a
short, high temperature treatment prior to a prolonged
treatment at a lower temperature, thereby reducing pain
and discomfort. Since temperatures >43 8C cannot be tolerated systemically, a local heat boost would be required to
take advantage of this effect for whole body hyperthermia. So
far, there is no evidence that tumor cells are differently
sensitized by step-down heating than normal cells.
It was initially thought that tumor cells have intrinsically

higher heat sensitivity than normal cells, but this is not
universally true. Although some neoplastic cells are more
sensitive to heat than their normal counterparts, this
appears to be the case at temperatures higher than those
used in systemic hyperthermia. Tumors in vivo, on the
other hand, often do have a higher thermal sensitivity than
normal tissues because of abnormal vasculature (reduced
blood flow), anaerobic metabolism (acidosis), and nutrient
depletion. Due to their tortuous and poorly constructed
vasculature, tumors have poor perfusion, thus heat dissipation by convection is reduced. At high temperatures
(43 8C and up) this means that tumors become a heat
reservoir with a consequent rise in temperature, which
if maintained for too long damages the microcirculation
and further impairs convective heat loss. Also increased
fibrinogen deposition at damaged sites in the vascular wall
leads to clusion of tumor microvessels. Significant heating
of the tumor cells results, which may be directly cytotoxic.
Additionally, the impaired blood flow brings about acidosis,
increased hypoxia and energy depletion all of which
increase the heat sensitivity of tumor cells (8). At lower
temperatures, typical of those achieved in whole-body
hyperthermia, blood flow increases (9) though the mechanism is not well understood. For these reasons, along with
the historical evidence for antitumor effects of fever and the
metastatic nature of malignant disease, cancer has become
the main focus of systemic hyperthermia.
Systemic hyperthermia results in increased delivery of
drugs to tumor sites because of increased systemic blood
flow. It can also increase blood vessel permeability by
increasing the effective pore size between the loosely bound
endothelial cells forming tumor microvessels, permitting
larger molecules, such as nanoparticles and gene therapy
vectors, to pass into the interstitium (10). Figure 3 shows
increased uptake of 210 nm liposomes in rat breast tumors
after 1 h of 41.5 8C whole-body hyperthermia. Heat may
also be toxic to endothelial cells, resulting in a transient
normalization of vascular architecture and improvement
in blood flow (11). Another barrier to drug delivery is the
high interstitial pressure of many tumors. Since wholebody hyperthermia, even at fever-range temperatures,
causes cell death (apoptosis and necrosis) within tumors
it reduces the oncotic pressure allowing greater penetration
of large molecules. Table 3 summarizes the interactions of
systemic hyperthermia which facilitate nanoparticle delivery to tumors.
Table 3. Whole-Body Hyperthermia Facilitates Nanoparticle Therapy

Heat Interaction
Therapeutic Effect
" Blood flow

" In endothelial gap size
" Endothelial cell apoptosis/necrosis ! transient normalization of vasculature
" Tumor cell apoptosis/necrosis # oncotic pressure
Temperature-dependent " in permeability of liposome bilayer
Cellular and molecular effects in tumor
Direct interactions with drug
"
"
"
"
"
"
"
Nanoparticle delivery to tumor

Nanoparticles in interstitium
And synchronization of drug release
Drug in tumor cell " drug efficacy
Drug efficacy
50
Whole-Body Hyperthermia and the Immune System

An increase in ambient temperature can serve as a natural
trigger to the immune system and it appears that the thermal
microenvironment plays a critical role in regulating events in
the immune response. The early work of Coley on cancer
therapy with infectious pyrogens implicated fever-induced
immune stimulation as the mediator of tumor responses (1).
While there have been numerous in vitro studies of the effect
of temperature on components of the immune system, indicating that the thermal milieu regulates T lymphocytes,
natural killer (NK) cells, and dendritic cells (DC), in vivo
examinations of the immune effects of systemic hyperthermia are relatively few. Initial animal model studies concluded that whole-body hyperthermia resulted in
immunosuppression, but high temperatures were used,
tumors were mostly immunogenic, and immune response
was merely inferred from the incidence of metastatic spread
rather than from measurement of specific markers of
immune system activation. The majority of in vivo studies
in animals provide evidence of a nonspecific host reaction
in response to hyperthermia in which both T and B
lymphocytes, as well as macrophages, are involved (12).
Although NK cells are intrinsically more sensitive in vitro
to heat than B and T cells, their activation by systemic
hyperthermia has been observed. Microwave induced
whole-body hyperthermia of unrestrained, unanesthetized mice at 39.540 8C for 30 min, three or six times
weekly, resulted in increased NK cell activity and reduced
pulmonary metastasis in tumor-bearing mice, but none in
normal mice (13). Evidence for hyperthermia-induced
human tumor lysis by IL-2 stimulated NK cells activated
by HSP72 expression also exists (14). Increased numbers
of lymphocyte-like cells, macrophages, and granulocytes
are observed in the tumor vasculature and in the tumor
stroma of xenografts and syngeneic tumors in mice immediately following a mild hyperthermia exposure for 68 h.
In the SCID mouse/human tumor system tumor cell
apoptosis seen following treatment was due largely to
the activity of NK cells. The investigators hypothesize
heat dilatation of blood vessels and increased vessel permeability may also give immune effector cells greater
access to the interior of tumors (15). In balb/C mice,
fever-range whole-body hyperthermia increased lymphocyte trafficking, resulting in early responsiveness to antigen challenge (16). Thus systemic hyperthermia may be
an effective, nontoxic adjuvant to immunotherapy.
A recent clinical study examined the effect of wholebody hyperthermia combined with chemotherapy on the
expression up to 48 h later of a broad range of activation
markers on peripheral blood lymphocytes, as well as serum
cytokines and intracellular cytokine levels in T cells, and
the capacity of these cells to proliferate. Immediately after
treatment with 60 min of 41.8 8C WBH as an adjunct to
chemotherapy, a drastic but transient, increase in peripheral NK cells and CD56 cytotoxic T lymphocytes was
observed in the patients peripheral blood. The number
of T cells then briefly dropped below baseline levels, a
phenomeonon that has also been observed by others (17).
A marked, but short-lived, increase in the patients serum
levels of interleukin-6 (IL-6) was also noted. Significantly
increased serum levels of tumor necrosis factor-alpha

(TNF-alpha) were found at 0, 3, 5 and 24 h posttreatment.
Further immunological consequences of the treatment
consisted of an increase in the percentage of peripheral
cytotoxic T lymphocytes expressing CD56, reaching a maximum at 48 h post-WBH. Furthermore, the percentage of
CD4+ T cells expressing the T cell activation marker CD69
increased nearly twofold over time, reaching its maximum
at 48 h. Since similar changes were not observed in
patients receiving chemotherapy alone, this study provided strong evidence for prolonged activation of human
T cells induced by whole-body hyperthermia combined with
chemotherapy (18).
Activation of monocytes has been observed following hot
water bath immersion such that response to endotoxin
stimulation is enhanced with concomitant release of
TNF-a. Macrophage activation and subsequent lysosomal
exocytosis were observed in the case of a patient treated for
liver metastases by hyperthermia. Lysosomal exocytosis
induced by heat may be an important basic reaction of the
body against bacteria, viruses, and tumor growth and was
proposed as a new mechanism of thermally induced tumor
cell death mediated by an immune reaction (19).
Several investigators have suggested that the immune
changes seen during in vivo whole-body hyperthermia are
mediated by elevations in the plasma concentrations of
either catecholamines, growth hormone, or beta-endorphins.
In volunteers immersed in a heated water bath, neither
recruitment of NK cells to the blood, nor the percentages
or concentrations of any other subpopulations of blood mononuclear cells were altered by hormone blockade. However,
somatostatin partly abolished the hyperthermia induced
increase in neutrophil number. Based on these data and
previous results showing that growth hormone infusion
increases the concentration of neutrophils in the blood, it
was suggested that growth hormone is at least partly
responsible for hyperthermia induced neutrophil increase.
A similar study suggested that hyperthermic induction
of T lymphocytes and NK cells is due to increased secretion of somatotropic hormone (7).
The peripheral blood level of prostaglandin E2 (PGE2),
which may act as an angiogenic switch, transforming a
localized tumor into an invasive one by stimulating new
blood vessel growth, and which also has an immunosuppressive effect, is elevated in patients with tumors compared to healthy control subjects. In a clinical study of
cancer patients receiving 12 h of 41.842.5 8C whole-body
hyperthermia, or extracorporeal hyperthermia, blood
levels of PGE2 decreased markedly after treatment and
correlated with tumor response (20).
In addition to their role as protectors of unfolding
proteins, extracellular heat shock proteins (HSP) can act
simultaneously as a source of antigen due to their ability to
chaperone peptides and as a maturation signal for dendritic cells, thereby inducing dendritic cells to cross-present
antigens to CD8 T cells (21). Heat shock proteins can also
act independently from associated peptides, stimulating
the innate immune system by eliciting potent proinflammatory responses in innate immune cells. The heat shock
response also inhibits cyclooxygenase-2 gene expression at
the transcriptional level by preventing the activation of
nuclear factor-kappaB (NFkB) (22). Thermal upregulation

of HSPs (HSP70 and HSP110) is strongest in lymphoid
tissues and may relate to the enhanced immune responses
that are observed during febrile temperatures. It has been
proposed that local necrosis induced by hyperthermic
treatment induces the release of HSPs, followed by uptake,
processing and presentation of associated peptides by dendritic cells. By acting as chaperones and as a signal for
dendritic cell maturation, HSP70 might efficiently prime
circulating T cells. Therefore, upregulating HSP70 and
causing local necrosis in tumor tissue by hyperthermia
offers great potential as a new approach to directly activate
the immune system, as well as to enhance other immunotherapies (23,24).
CLINICAL TOXICITIES OF WHOLE-BODY HYPERTHERMIA

TREATMENT
At fever-range temperatures, adverse effects of systemic
hyperthermia treatment are minimal however, at higher
temperatures they can be significant, even fatal. On the
other hand, the teratogenic effects (birth defects, still
births, spontaneous abortions) and 8Cular damage (cataract induction) resulting from electromagnetic fields used in
local hyperthermia are not seen in systemic hyperthermia.
The transient cardiorespiratory effects of elevated temperature can, however, lead to severe toxicity. Elevated
heart rate, especially at high temperatures may result in
arrythmias or ischemic heart failure, consequently
patients have to be very carefully screened with regard
to their cardiac status. Beta blockade has generally been
found to be deleterious although infusion of esmolol has
been safely carried out (25). Pulmonary hypertension and
edema due to capillary leak may also be seen, but like the
cardiac effects, these return to baseline a few hours after
treatment. Increased serum hepatic enzymes have been
noted, but these may be cancer related. All these toxicities
are less prevalent or less severe with radiant heat systems,
particularly at lower temperatures, and when light conscious sedation is used rather than general anesthesia. For
example, decreased platelet count, decreased plasma fibrinogen, and other factors leading to increased blood clotting
have been noted, particularly in extra-corporeal hyperthermia, but also with other methods of heating carried
out under inhalation-administered anesthesia drugs.
On the other hand, with whole-body hyperthermia under
conscious sedation there is no evidence of platelet
drops (26) and animal studies even show platelet stimulation providing protection against radiation induced
thrombocytopenia.
Since systemic hyperthermia is almost never used as a
single treatment modality, it is important to recognize that
whole-body hyperthermia combined with radiation and
chemotherapy can enhance some of the toxicities associated with these modalities. For example, the cardiotoxicity of doxorubicin and both the renal toxicity and
hematological toxicity of platinum agents may increase
under hyperthermia (27), while the muscle and peripheral
nervous system effects of radiation and some drugs can also
be enhanced (28). Bone marrow suppression is the limiting
51
toxicity of many chemotherapy drugs but there is little data

to suggest that whole body hyperthermia exacerbates this
effect. On the contrary, the synergy of hyperthermia with
several chemotherapy agents may mean that lower doses
can be used, resulting in less toxicity. For example, systemic hyperthermia combined with carboplatin achieves
therapeutic results without elevation of myelosuppression
and responses have occurred at lower than normal doses
(29). Pressure sores can easily develop at elevated
temperatures thus care must be taken not only in patient
placement and support, but also with application of monitoring devices. If heat dissipation is locally impaired, for
example, at pressure points, hot spots occur that can lead to
burns. This is rarely a problem with fever-range wholebody hyperthermia, but in anesthetized patients undergoing high heat regimens burns are not uncommon.
Following systemic hyperthermia treatments, malaise
and lethargy are almost universally experienced although
these may be counteracted by pain relief and a sense of
well-being due to released endorphins. However, the faster
the target temperature is reached, the less the exhaustion
(6), thus attention to minimizing heat dissipation during
the heat-up phase and using efficient heating devices, such
as those that generate heat by several mechanisms (e.g.,
radiant heat and EM fields), add a regional heat boost, or
produce near-IR radiation that is preferentially absorbed,
is advantageous to patient well being. Fever after treatment in the absence of infectious disease is not uncommon
and may be associated with an inflammatory response to
tumor regression. Nausea and vomiting during the first
couple of days after treatment are also common. Outbreaks
of herpes simplex (cold sores) in susceptible individuals
have also been noted, but are easily resolved with acyclovir.
THERMAL DOSE
The definition of dose for systemic hyperthermia is problematic. An applied dose would be the amount of heat energy
generated or delivered to the body but even if it can be
measured, this quantity does not predict biological effects.
By analogy with ionizing radiation, the absorbed dose
would be amount of thermal energy absorbed per unit
mass of tissue (J
kg1), however, this is not a quantity
that can be readily measured, or controlled, neither would
it necessarily predict biological effects. As indicated in the
previous sections, the effects of systemic hyperthermia
depend on (1) the temperature, and (2) the duration of
heating, but not on the energy required to produce the
temperature rise. This leads to the concept of time at a
given temperature as a practical measure of dose. In
reality, however, temperature is seldom constant throughout a treatment, even in the plateau phase of systemic
hyperthermia, so time at temperature is at best a crude
measure. Nonetheless, it is the one that is used most often
clinically for whole-body hyperthermia because of its simplicity. Ideally, the dose parameter should allow for comparison of treatments at different temperatures. Based on
the Arrhenius relationship and measured cell growth inhibition curves, the heating time at a given temperature
relative to the heating time at a standard temperature or
52
thermal dose equivalent (TDE), was defined empirically as,

T1 t2
RT1 T2
10
A discontinuity occurs in the temperature-time curves

between 42 and 43 8C for both cells in culture and heated
tissues, thus the value of R changes for temperatures above
the transition: R 2 < 42.5 8C and R 5 > 42.5 8C in vitro
while for in vivo heating studies, R 2.1 below the transition temperature and 6.4 above 42.5 8C. In practice, a finite
time is required for the body or tissue of interest to reach the
target temperature, temperature fluctuates even after the
target temperature is reached, and there is a cooling period
after heating ceases. If the temperature is measured frequently throughout treatment, the temperaturetime
curves can be integrated to provide the accumulated thermal dose that produces an equivalent effect to that resulting
from holding the cellstissue at a constant reference temperature for a given a period of time:
t43
Zt f
R43Tt dt
11
ti
where ti and tf are the initial and final times of the heating
procedure (30). This thermal isoeffect dose (TID) is usually
expressed in minutes is sometimes known as the tdm43 or
the cumulative equivalent minutes (CEM 43 8C). While a
biological factor has now been built in to the dose measure,
and the integrated TID allows for temperature variations
during heat-up and cool-down phases, it does not take into
account thermal tolerance and step-down sensitization. Nor
is it particularly relevant to clinical whole-body hyperthermia where multiple physical and biological effects combine
in a complex manner although for a given patient, time
temperature profiles are generally reproducible from one
treatment to another. A further modification attempts to
take into account temperature inhomogeneity through the
measurement of temperature at multiple sites and defining
T90, namely, that temperature exceeded by 90% of the
measurements (or correspondingly 20%: T20; or 50%:
T50). The TID is then expressed as cumulative equivalent
minutes that T90 is equal to 43 8C (CEM 43 8C T90) (31).
The efficiency of adjuvant hyperthermia in enhancing
the biological effectiveness of other treatments is often
reported in terms of the thermal enhancement factor
(TEF) or thermal enhancement ratio (TER). This quantity
is defined in terms of the isoeffect dose as,
dose of treatment to achieve
a given endpoint
TER
dose of treatment with heat
to achieve the same endpoint
12
In clinical and laboratory studies, the TER is often

computed on the basis of isodose rather than isoeffect,
for example, in the case of hyperthermia plus drug
induced arrest of tumor growth, TER TGDHT/TGTRT,
where TGD HT is the tumor growth delay due to
hyperthermia plus chemotherapy, and TGTRT is the
tumor growth delay resulting from chemotherapy at room
temperature. Similarly, the enhancing effect of hyperther-
mia on radiation treatment may be expressed through

TER D0HT/D0RT or TER LD50HT/LD50RT, where D0
is the time required to reduce survival to 1/e of its initial
value, and LD50 is the lethal dose to 50% of cells.
TEMPERATURE MEASUREMENT
Since systemic hyperthermia achieves a uniform temperature distribution, except for possible partial sanctuary sites,
thermometry for systemic hyperthermia is much less challenging than for regional or intracavitary hyperthermia, but
it is still important to prevent adverse effects, especially
burns. Also, convection can induce steep thermal gradients,
especially around major blood vessels, so that careful placement of temperature probes is required. Most practitioners
of whole-body hyperthermia measure temperature in several locations, typically the rectum, the esophagus, and at
several skin sites. During heat-up, the esophageal temperature is usually 12 8C higher than the rectal temperature,
but during plateau phase it drops to 0.51.5 8C below the
rectal temperature. Continuous and accurate temperature
measurement is particularly important when temperatures
>418C are to be achieved, as critical, life-threatening
changes can occur in minutes or even seconds and over
changes in temperature of as little as 0.10.2 8C because of
the nonlinear response to temperature. For moderate temperature systemic hyperthermia, temperature measurement to within 0.1 8C is usually adequate, but a precision
of 0.01 8C is desirable when heating to >41 8C and also
allows determination of the specific absorption rate from the
slope of the temperature versus time curve. The temperature measuring device must be insensitive to all other
influences, such as ambient temperature, moisture, nearby
electromagnetic fields, and so on and satisfying this criterion can be difficult. Frequent calibration of thermometers in
the working range of temperatures is important since some
thermometers appear fine at 30 8C, but drift substantially at
40 8C and above. Stringent quality control of any thermometry system is required to monitor accuracy, precision,
stability, and response time.
Table 4 summarizes the different types of thermometer
probes available for internal and external body temperature measurements, and their relative merits and disadvantages for systemic hyperthermia. Thermistors are most
often used for standard temperature monitoring sites while
thermocouples are used for tumor or other intra-tissue
measurements. Recently, noninvasive methods of temperature measurement have been developed that are
beginning to see application in hyperthermia. Thermography provides a two-dimensional (2D) map of surface temperature by measurement of infrared emission from the
body, though deep-seated hot structures may be visualized
because of heat carried by blood flow from the interior heat
source to the skin. It is useful to detect skin hotspots and
therefore in burn prevention. Since temperature-induced
changes in the mechanical properties of tissue lead to
altered ultrasound propagation velocity, mapping of ultrasound velocity can also provide a visual map of temperature. Tomographic reconstruction of 2D or 3D temperature
is theoretically possible, but it is difficult in practice
because of the heterogeneity of tissue characteristics. A
53
Thermistor (e.g.,
Bowman
Loop Larsen probe)
GaAs
Optical (fiber optic probe):
LCD birefringent crystal
fluorescent phosphor
Thermocouple
Refraction of polarized light

Decay of fluorescence
Seebeck effect: temperature

dependent voltage
difference between two
conductors made of different
metals
Inverse relationship between
temperature
and semiconductor resistance
Temperature specific absorption
Change w/temp.:
Color reflectance
Expansion of mercury or
alcohol in glass
Linear resistance change
with temperature
Clinical
Platinum resistance
thermometer
Measurement Principle
Probe Type
Table 4. Temperatures Probes for Systemic Hyperthermia
Very high
Low
Moderate
Low
Low
High
High < 0.05 8C
Moderate 0.1 8C
Moderate to
high
Low
Moderate 0.1 8C
High 0.02 8C
Sensitivity
Accuracy
Low
Poor Require frequent

recalibration
Used as standard for

calibration
of other types
of thermometers.
Moderate
High
Stability
Small size
Not sensitive to EM fields. Small size.
Unstable
Short time constant. Not interchangeable.

Sensitive to EM fields.
Small sensor. Nonlinear voltage

change with temp. Sensitive to EM
fields. Cant handle steep temp. gradients.
Expensive. Difficult to calibrate. Large size.

Sensitive to shock.
Large size, inflexible. Slow response.
Advantages or Disadvantages
54
2005
2002
2004
Hegewisch-Becker, S.
Hildebrandt, B.
2004
Bull, J.M.
Guan, J.
2002
Bull, J.M.
1990
1992
Bull, J.M.
Engelhardt, R.
Phase II
WBH Chemotherapy
Bakshandeh, A.
2003
2004
Phase II
1994
Steinhausen, D.
Douwes, F.
Phase I
2002
WBH Alone
Kraybill, W.G.
Phase I/II
Phase II
Phase II
Pilot
Pilot
Phase I
Phase I
Phase I
Public Study
Year
Type
First Author
28
41
32
23
21
33
13
17
25
103
Pretreated advanced
metastatic
colorectal cancer
Metastatic colorectal
cancer
Advanced cancers
Advance metastatic
melanoma
Advanced metastatic
cancers (GI, breast,
head and neck, sarcoma,
neuroendocrine)
Ovarian cancer
Various chemotherapy
resistant cancers
Advanced metastatic
sarcoma
Nonmetastatic malignant
pleural mesothelioma
Advanced refractory or
recurrent cancers
Advanced solid tumors
Number of Disease
Patients
Table 5. Summary of Clinical Trials of Whole-Body Hyperthermiaa
1 h at 41.842.1 8C
hyperglycemia
hyperoxemia
folinic acid 5-FU
mitomycin C
1 h at 41.8 oxaliplatin
leucovorin 5FU
1-2 h at 41.542.0 8C
cisplatin
or carboplatin
hyperglycemia
1 h at 41.0 8C cisplatin
doxorubicin
6 h at 40.0 8C cisplatin
gemcitabine
interferon-a
6 h at 40.0 8C doxil 5-FU

metronomic interferon-a
2 h at 41.842.0 8C BCNU
1 h at 41.8 8C ifosfamide
carboplatin etoposide
1 h at 41.8 8C
hyperglycemia
hyperoxemia
36 h at 39.540.0 8C
Protocol
Grade III/IV toxicities 11

responses/SD (39%)
52, 2198312
Slight " in myelotoxicity 10

responses/SD Response rate
that in literature for
chemo alone
94% responses/SD Pain
reduction in all pts.
Increased KPS
Decreased tumor markers
No excess toxicity 31
responses/SD (76%)
44, 15204528
55, 12181242
65
15108039
35
60
33, 1607734
12609573
8023241
16, 12028640
Reference,
PMID
18 responses/SD (86%) "

quality of life
Grade III/IV neutropenia and

thrombocytopenia 5 partial
remissions (20%)
Limiting toxicity
thrombocytopenia
7 responses/SD (41%)
" survival
Grade III toxicities
9 responses/SD
(69%)
20 responses/SD (66%) "
survival " quality of life
Well tolerated No significant

adverse events # in circulating
lymphocytes
Minimal side effects 52
responses (50%)
Result of WBH
55
2005
2004
1993
1997
2004
2001
2003
1995
1990
Kurpeshev, O.K.
Richel, O.
Robins, H.I.
Robins, H.I.
Strobl, B.
Westermann A.M.
Westermann A.M.
WBH Radiation
Overgaard, J.
Robins, H.I.
Published since 1990.
2005
Ismael-Zade, R.S.
2004
Hou, K.
95
14
30
16
21
42
54
Randomized
70
multicenter
Pilot
8
Phase II
Phase II
Phase II
Phase I
Phase I
Phase II
Phase II
Pilot
Phase II
1 h at 41.8 8C ifosfamide
carboplatin etoposide
1 h at 41.8 8C carboplatin
1 h at 41.541.8 8C paclitaxel
carboplatin
1 h at 41.8 8C L-PAM
3 h at 41.842.5 8C
doxorubicin
interferon-a
12 h at 41.042.3 8C
poly-chemotherapy
12 h at 41.842.5 8C,
extracorporeal
chemotherapy
vs. chemotherapy alone
1 h at 43 8C fractionated RT
vs. FRT alone
Nodular lymphoma, chronic 41.8 8C TBI vs. LON TBI
lymphocytic leukemia
Metastatic melanoma
Metastatic sarcoma
Platinum resistant
ovarian cancer
Metastatic cervical cancer
Various refractory cancers

Various refractory cancers
Various disseminated
cancers
Metastatic and recurrent
cervical cancer
Pediatric renal cell

carcinoma
Advanced cancers
Improved local tumor control,

" survival
8 responses/SD " survival
Regression of metastases.
Pain reduction
Grade III/IV leucopenia,
thrombopenia,
anemia, renal toxicity 16
responses/SD (76%)
Myelotoxicity 9 responses (30%)
Lower platelet nadir
myelosuppression 8
responses/SD (50%)
Grade II alopecia Grade III/IV
thrombopenia, neutropenia
" survival
Grade IV thrombocytopenia,
grade III neutropenia 9
responses/SD (64%)
Neutropenia, thrombocytopenia,
infection 58 responses/SD (61%)
24, 2182581
41, 7776772
50, 12759526
11378341
58
53, 8355046
48, 8996137
57, 15581981
66
68
75.3% responses/SD 72.6% #
tumor markers 70% pain
relief improved sleep " weight,
appetite, KPS All signifly > control
Reversible toxicities
No complications 5 responses (100%) 15700247
number of magnetic resonance (MR) techniques have been

used for thermal mapping and BSD Medical and SIEMENS
Medical Systems have collaborated to develop a hybrid
hyperthermia/MRI system, although it is not a whole-body
hyperthermia machine. Currently, the most widely
accepted MR technique is the proton resonance frequency
(PRF) method that exploits the temperature dependence of
the chemical shift of water. Unlike the value of the water
spin-lattice relaxation time or the molecular diffusion
coefficient, both of which have been used for MRI temperature measurements, the thermal coefficient relating temperature to the water chemical shift has been shown to be
essentially independent of tissue type and physiological
changes induced by temperature (32). Recently an interleaved gradient echoecho planar imaging (iGE-EPI)
method for rapid, multiplanar temperature imaging was
introduced that provided increased temperature contrastto-noise and lipid suppression without compromising spatio-temporal resolution (33).
CLINICAL EXPERIENCE
Cancer
Systemic hyperthermia has been used mostly for treatment
of cancer because of its potential to treat metastatic disease. Initial treatments aimed to produce direct killing of
tumor cells based on the premise, now understood not to be
universally true, that cancer cells are more susceptible to
elevated temperatures than normal cells, and the higher
the temperature the greater the tumor cell kill. Maximally
tolerated temperatures of 41.542 8C were therefore maintained for 12 h as the sole treatment. Response rates were,
however, disappointing. Tumor regressions were observed
in less than half the cases, no tumor cures were achieved,
and remissions were of short duration. It became apparent
that the heterogeneity of cell populations within tumors,
along with micro-environmental factors, such as blood/
nutrient supply, pH, and oxygen tension prevent the thermotoxic results achieved in the laboratory. Consequently,
the focus of research on systemic hyperthermia shifted to
using hyperthermia as an adjunct to other cancer therapies, principally chemotherapy and radiotherapy. It is
important to note that because of the experimental status
of systemic hyperthermia treatment for cancer, almost all
clinical trials, summarized in Table 5, have been performed
on patients with advanced disease for whom whole-body
hyperthermia, either as a sole therapy, or as an adjunct, is
a treatment of last resort. In these cases, any response
whatsoever is often remarkable. Nonetheless, a number of
hyperthermia centers in Europe have discontinued systemic hyperthermia because the high temperature protocols required intensive patient care and led to unacceptable
toxicities, especially in light of the efficacy and reduced
toxicities of newer generation chemotherapies. Large, randomized, multicenter, Phase III trials are, however, needed
to firmly establish the benefits of systemic hyperthermia in
conjunction with chemotherapy and radiation. Also, validation and optimization of fever-range temperature protocols are much needed.
Systemic Hyperthermia and Chemotherapy. The beneficial interaction of hyperthermia with several classes of
chemotherapy agents, acting via several mechanisms as
summarized in Table 6, has spurred a variety of thermochemotherapy regimens and several clinical trials of systemic hyperthermia and chemotherapy are ongoing. While
the results have been mixed, elevated response rates were
recorded in the treatment of sarcoma when systemic
hyperthermia was combined with doxorubicin and cyclophosphamide (54) or BCNU (34). Systemic hyperthermia is
the only way to heat the lung uniformly, and impressive
response rates and increased durations of response have
been achieved in both small cell and nonsmall cell lung
cancer treated with the combination of whole body
hyperthermia at 41 8C for 1 h with adriamycin, cyclophosphamide, and vincristine (ACO protocol) (34). Neuroendocrine tumors also appear to have increased sensitivity to
systemic hyperthermia and multidrug chemotherapy (51).
Optimal combination of whole-body hyperthermia with
chemotherapy requires an understanding of the mechanisms of interaction of heat with individual drugs or drugs in
combination. Preclinical data is consistent with the concept
that the timing of chemotherapy during whole-body
hyperthermia should affect therapeutic index. For example, Fig. 4 shows the effect on tumor cures in mammary
carcinoma bearing rats of 6 h of 40 8C whole-body
hyperthermia administered with, or 24 or 48 h after gemcitabine. A synergistic response was obtained when
hyperthermia was begun with gemcitabine administration
or 48 h later. The effect of gemcitabine was completely
negated, however, when hyperthermia was administered
24 h after the start of heating, perhaps due to cell cycle
effects. With cisplatin, the greatest therapeutic index is
achieved if the drug is given 24 h before the start of wholebody hyperthermia, thereby preventing thermal augmentation of cisplatin induced nephrotoxicity (55). In a clinical
investigation of multiple cycles of radiant heat whole-body
hyperthermia combined with carboplatin, Ifosfamide, etoposide, and granulocyte colony stimulating factor, it was
found that toxicity was minimized when carboplatin was
GEM+WBH(0)
TER for tumor cures
56
GEM+WBH(+24)
GEM+WBH(+48)
5.0
4.0
3.57
2.98
3.0
2.0
1.0
0.0
0.06
GEM+WBH(0 h) GEM+WBH(+24 h) GEM+WBH(+48 h)
GEM+WBH schedule
Figure 4. Schedule dependence of fever range whole-body
hyperthermia enhanced gemcitabine tumor cures. A supraadditive cure rate occurred when whole-body hyperthermia (WBH)
was given at the same time as gemcitabine administration or 48 h
later. When hyperthermia followed gemcitabine by 24 h the number
of cures dropped to almost zero, well below the number achieved with
gemcitabine alone.
57
Table 6. Chemotherapy Agents Used with Whole-Body Hyperthermia

Class of Agent
Alkylating agents
Likely Mechanism
of Heat Interaction
Drugs Used with

WBH in Clinical Studies
Investigator
Referencesa
Impaired DNA repair

Improved pharmacokinetics
Cyclophosphamide (CTX)
Parks, 1983 (4)

Engelhardt, 1988 (34)
Lange, 1983 (35)
Robins, 1997 (36)
Issels, 1990 (37)
Westermann, 2003 (38)
Parks, 1983 (4)
Bull, 1992 (39)
Bull, 1992 (39)
Parks, 1983 (4)
Herman, 1982 (40)
Robins, 1993 (42)
Douwes, 2004 (43)
Hegewisch-Becker, 2002
Douwes, 2004 (43)
Richel, 2004 (46)
Strobl, 2004 (47)
Elias, 2004 (48)
Bull, 2002 (49)
Herman, 1982 (40)
Lange, 1983 (35)
Larkin, 1979 (50)
Bull, 2002 (49)
Bull, 2004 (51)
Barlogie, 1979 (52)
Issels, 1990 (37)
Elias, 2004 (48)
Strobl, 2004 (49)
Strobl, 2004 (49)
Robins, 1989 (53)
Dacarbazine (DTIC)
Melphalan (L-PAM)
Ifosfamide (IFO)
Nitrosoureas
Impaired DNA repair

Improved pharmacokinetics
Platinum agents
Impaired DNA repair

Altered plasma protein binding
BCNU
Me-CCNU
Cisplatin (CDDP)
Carboplatin (CBDCA)
Oxaliplatin
Anthracyline antibiotics
Impaired DNA repair

Enzyme activation
Antimetabolites
Increased drug transport

Cell cycle arrest
Impaired DNA repair
Adriamycin
Bleomycin
5-FU
Antiproliferatives
Impaired DNA repair
Gemcitabine
Etoposide (VP-16)
Topoisomerase inhibitors
Impaired DNA repair
Irinotecan (CPT-11)
Taxanes
Microtubule disruption
Apoptosis
Increased anti-viral and
antiproliferative activity
Paclitacel
Docetaxel
Interferon
Biological response
modifiers
(44)
(45)
(44)
(44)
(45)
Bull, 2002, 2004 (34,49)

a
References prior to 1980, or not in English, are not provided in the Bibliography at the end of this article.
given during the plateau phase of WBH, 10 min after target

temperature was reached (56).
A major rationale for whole-body hyperthermia in cancer treatment is the ability to treat metastases, but this is
actually a controversial issue. There have been no clinical
studies specifically designed to study the effect of systemic
hyperthermia on either the efficacy against metastatic
disease or prevention of development of metastases.
Increased survival in advanced malignancies is often interpreted to mean a reduction in metastatic disease, but direct
measurement of the incidence and response of metastases
is rare. Based on some animal studies, it has been suggested that systemic hyperthermia could actually promote
the metastatic spread of tumor cells, but this has not been
confirmed. One clinical study found an increase of tumor
cells in blood 24 h after 41.8 8C WBH, but there was no
evidence that this caused metastatic spread of disease (57).

Several animal experiments do support the efficacy of
whole-body hyperthermia against metastases. In mouse
models of lung cancer and melanoma, the number of lung
metastases was scored after repeated systemic microwave
heating. It was found that the number of lung metastases
was significantly reduced, and NK-cell activity was higher,
in treated animals. The authors hypothesized that WBH
interferes with the spread of organ metastases, possibly
through a mechanism involving NK cells (13). Another
study of mouse Lewis lung carcinoma in which the animals
were treated with 60 min of systemic hyperthermia at
42 8C, demonstrated a reduction in the number and
percentage of large metastases (>3 mm) on day 20 posttumor implantation. Addition of radiation led to a reduction to 50% of control of the number of lung metastases as
58
well as the percent of large metastases on day 20 (58). In a

breast cancer ocult metastasis model in rats, 6 h of 40 8C
whole-body hyperthermia combined with daily, low dose,
metronomic irinotecan resulted in delayed onset, and
reduced incidence, of axillary lymph node metastases compared to control in rats, as did treatment with 40 8C WBH
alone. The combination therapy also reduced axillary
metastasis volume. Interestingly, none of the therapies
significantly affected inguinal lymph node metastases,
but lung metastases were decreased in both the combination therapy and WBH alone groups. Rats treated with
fever-range whole-body hyperthermia and metronomic
irinotecan also survived significantly longer (36%) than
control animals (59).
Systemic Hyperthermia and Radiotherapy. The augmentation of ionizing radiation induced tumor kill by
hyperthermia is well documented for local hyperthermia
and has led to numerous protocols combining whole-body
hyperthermia with radiation therapy (60,61). Hyperthermia is complementary to radiation in several regards:
ionizing radiation acts predominantly in the M and G1
phases of the cell cycle while hyperthermia acts largely
in S phase; radiation is most effective in alkaline tissues
whereas hyperthermic cytotoxicity is enhanced under
acidic conditions; radiation is not effective in hypoxic
regions yet hyperthermia is most toxic to hypoxic cells.
Thus when hyperthermia is combined with radiotherapy,
both the hypoxic, low pH core of the tumor is treated as well
as the relatively well perfused outer layers of the tumor.
Furthermore, because of its vascular effects, hyperthermia
enhances tumor oxygenation thus potentiating radiation
cell kill. Hyperthermia also increases the production of oxygen radicals by radiation, and reduces the repair of DNA
damage caused by ionizing radiation. Thus hyperthermia
and radiotherapy together often have a synergistic effect,
and this combination is now well accepted for treatment of a
number of tumors.
Fever-Range WBH. Like systemic hyperthermia alone,
combined modality treatments were initially aimed to
achieve maximally tolerated temperatures. Such regimens, however, carry significant risk to the patient,
require general anesthesia, and necessitate experienced,
specialist personnel to provide careful monitoring of vital
signs and patient care during the treatment. More
recently, it has been appreciated that lower core body
temperatures (3940 8C) maintained for a longer time
(48 h), much like fever, can indirectly result in tumor
regression through effects on tumor vasculature, the
immune response, and waste removal (detoxification).
The optimum duration and frequency of mild hyperthermia treatment has, however, not yet been determined.
Protocols range from single treatments of 46 h, or similar
long duration treatments given once during each cycle of
chemotherapy, to daily treatments of only 1 h. Several
studies of mild, fever-range, whole-body hyperthermia
with chemotherapy have demonstrated efficacy against a
broad range of cancers (34,17) and clinical trials are currently being conducted at the University of Texas Health
Science Center at Houston, Roswell Park Cancer Institute,
New York, and by the German Interdisciplinary Working

Group on Hyperthermia (62).
Systemic Hyperthermia and Metabolic Therapy.
Increased rates of metabolic reactions lead to rapid turnover of metabolites, causing cellular energy depletion,
acidosis, and consequent metabolic disregulation. Tumors,
which have increased metabolic rates [glucose, adenomine
triphosphate (ATP)] compared to normal cells, may be
particularly sensitive to thermally induced energy depletion and this has been exploited in the Cancer Multistep
Therapy developed by von Ardenne, which is a combined
hyperthermiachemotherapymetabolic therapy approach
to cancer (63). The core of this approach is systemic
hyperthermia at 4042 8C, sometimes with added local
hyperthermia to achieve high temperatures within the
tumor. A 10% solution of glucose is infused into the patient
to achieve a high accumulation of lactic acid within the
tumor that cannot be cleared because of sluggish blood flow
and confers an increased sensitivity to heat to the tumor
cells. Administration of oxygen increases the arterial oxygen pressure and stimulates lysozymal cytolysis. Finally
low dose chemotherapy is added.
Palliation. Pain relief is reported by many patients
receiving systemic hyperthermia treatment, whether with
chemotherapy or radiation. Indeed, almost all patients
undergoing thermoradiotherapy report pain relief.
Immediate pain relief following treatment is likely to stem
from an increased level of circulating b-endorphins, while
longer term pain relief may be due to increased blood flow,
direct neurological action, and disease resolution, for
example, tumor regression in cancer patients, or detoxification. Meaningful improvements in quality of life typically result from such pain relief. Localized infrared
therapy using lamps radiating at 225 mm is used for
the treatment and relief of pain in numerous medical
institutes in China and Japan.
Diseases Other than Cancer. Therapeutic use of heat
lamps emitting IR radiation is commonplace throughout
the Orient for rheumatic, neurological and musculoskeletal conditions, as well as skin diseases, wound healing, and
burns. The improvements reported appear to be largely
due to increased blood flow bringing nutrients to areas of
ischemia or blood vessel damage, and removing waste
products. Scientific reports of these treatments are, however, difficult to find. Application of heat via hot baths or
ultrasound has long been standard in physical therapy for
arthritis and musculoskeletal conditions, though ice packs
are also used to counter inflammatory responses. Heat
decreases stiffness in tendons and ligaments, relaxes the
muscles, decreases muscle spasm, and lessens pain. Unfortunately, few clinical trials of efficacy have been performed,
and methodological differences or lack of rigor in the
studies hinder comparisons (64). A clinical trial in
Japan reported a supposedly successful solution for seven
out of seven cases of rheumatoid arthritis treated with
whole-body IR therapy, and it is reported that the King of
Belgium was cured of his rheumatoid arthritis in three
months due IR treatments. Systemic hyperthermia with
THE FUTURE OF SYSTEMIC HYPERTHERMIA

While there is a resurgence of interest in systemic
hyperthermia, this modality has not yet been adopted as
a mainstream therapy, and optimal clinical trials have
not yet been carried out. Well-designed, well-controlled,
multicenter clinical trials need to be conducted. In order to
unequivocally demonstrate the utility of whole-body hyperthermia in the treatment of cancer as well as other diseases, it
will be important to accrue a sufficiently large number of
patients who do not have end-stage disease. Thanks to the
commercial availability of systemic hyperthermia systems,
the variability between induction techniques at different
institutions can be removed. Newer instrumentation, particularly near-IR radiant heat devices, along with treatment at
lower temperatures (fever-range thermal therapy) should
lead to significantly reduced toxicity. Better exploitation of
5
grp78-HSVtk
40C/37C ratio of HSVtk
molecules per cell
whole-body radiant heat units is being carried out in

clinical centers as well as many private clinics in Germany
for the purpose of alleviating rheumatoid arthritis. It has
been proposed that the induction of TNF receptors by WBH
may induce a remission in patients with active rheumatoid
arthritis. The use of heat packs has long been standard to
relieve the pain of fibromyalgia. Again, the therapeutic
effect is believed to be due to increased circulation flushing
out toxins and speeding the healing process. Whole-body
hyperthermia treatment for fibromyalgia and chronic fatigue syndrome (CFS) is to be found in a number of private
clinics. Hyperthermia increases the number and activity of
white blood cells, stimulating the depressed immune system of the CFS patient.
Because of its immune stimulating effects, whole-body
hyperthermia is a strong candidate for treatment of chronic
progressive viral infections, such as HIV and hepatitis C. A
clinical trial at the University Medical Center Utrecht, The
Netherlands has evaluated extracorporeal heating to induce
systemic hyperthermia of 41.8 8C for 120 min under propofol anesthesia for treatment of hepatitis C (65). Human
immunodeficiency virus (HIV)-infected T cells are more
sensitive to heat than healthy lymphocytes, and susceptibility increases when the cells are presensitized by exposure to tumor necrosis factor. Thus, induction of whole-body
hyperthermia or hyperthermia specifically limited to tissues
having a high viral load is a potential antiviral therapy for
acquired immunodeficiency syndrome (AIDS). An Italian
study has found treatment of AIDS with beta-carotene and
hyperthermia to be synergistic, preventing progression of
early disease and also increasing the survival time in
patients with severe AIDS. A single treatment of low flow
extracorporeal hyperthermia was found effective against
AIDS associated Kaposis sarcoma, though there was significant toxicity. Core temperature was raised to 42 8C and
held for 1 h with extracorporeal perfusion and ex vivo
blood heating to 49 8C. Complete or partial regressions
were seen in 20/29 of those treated at 30 days post-treatment, with regressions persisting in 14/29 of those treated
at 120 days post-treatment. At 360 days, 4/29 maintained
tumor regressions with 1 patient being in complete remission still at 26 months (66).
59
4
3
2
1
0
Liver
Tumor
Figure 5. Fever range whole-body hyperthermia increases

therapeutic gene (grp78-HSVtk) delivery in tumor.
the narrow window of effective temperatures within which

the cellular effects of heat can be exploited yet damage
remains minimal, and improved understanding of the biological interactions in vivo of systemic heat with chemotherapy
and radiation will be essential to optimize therapy.
The effects of systemic hyperthermia on tumor blood
flow and vascular permeability have the potential to
increase delivery of various small molecules, nanoparticles,
and gene therapy vectors to tumors. Ferromagnetic nanoparticles can be heated by external magnetic fields and offer
the potential for internal hyperthermia, both locally and
systemically. Thermally sensitive liposomes that release
their contents at designated temperatures are also of interest. The ability of systemic hyperthermia to aid in systemic
delivery of gene therapy vectors (the holy grail of gene
therapy) and enhance transfection of cells with therapeutic
gene plasmids is under investigation in several laboratories
(67,68), and shows potential along with targeted gene therapy via the heat shock response. For example, Fig. 5 shows a
fourfold hyperthermic increase of therapeutic gene delivery
to tumor when plasmid DNA was injected intravenously
into mammary carcinoma bearing rats immediately after 6
h of whole-body hyperthermia at 40 8C. Thus systemic
hyperthermia is likely to see increasing application as an
enhancer of drug delivery.
There is a great deal of interest in the immunological
consequences of whole-body hyperthermia, and as they
become better understood, the combination of systemic
hyperthermia with specific immunotherapies will undoubtedly be pioneered, not just for cancer but also, by analogy
with fever, in a broad range of diseases.
SUMMARY
Systemic hyperthermia is founded on solid physical and
biological principles and shows promise in the treatment
of a number of diseases. Modern whole-body hyperthermia devices use IR-A radiation sources together with
effective heat loss techniques to achieve a controlled,
uniform temperature distribution throughout the body
with minimal patient toxicity. A shift in paradigm has
occurred away from achieving direct cell killing with short
60
bouts of maximally tolerated temperatures, to inducing

indirect curative effects through longer duration treatments
at lower temperatures, and synergy with other modalities,
such as radiotherapy. Better understanding of the interactions of elevated temperature with metabolic and genetic
pathways will allow thermally driven targeted therapies.
Of particular promise is the use of systemic hyperthermia
as an immune system stimulator and adjunct to immunotherapy. Application of systemic hyperthermia to nanoparticle delivery and gene therapy is emerging. Whole-body
hyperthermia is moving from being dubbed an alternative
therapy to becoming a standard treatment and clinical
hyperthermia centers are to be found all over the world.
Useful Websites
http://www.ncbi.nlm.nih.
Techniques and
gov/books/bv.fcgi?rid=hstat6. Devices Used
section.40680
to Produce
Hyperthermia
http://www.ncbi.nlm.nih.gov/ Physics and
Physiology of
books/bv.fcgi?rid=cmed.
Heating
section.7813
Clinical Hyperthermia
http://www.duke.edu/~dr3/
Background
hyperthermia_general.
html
http://www.eurekah.com/isbn. Online book:
Locoregional
php?isbn=1-58706Radiofrequency-perfusional
248-8&bookid=143
and Whole Body
&catid=50
Hyperthermia in
Cancer Treatment:
New Clinical Aspects,
E.F. Baronzio and
A. Gramaglia (eds.),
Eurekah Bioscience
Database
http://www.esho.info/
European Society for
professionals/
Hyperthermic
Oncology
http://www.hyperthermie.org/ German
index2.html
Interdisciplinary
Working group on
hyperthermia
http://www.uth.tmc.edu/
Systemic Thermal
thermaltherapy/
Therapy at the
University of Texas
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
BIBLIOGRAPHY
1. Nauts HC. Bacterial pyrogens: beneficial effects on cancer
patients. In: Gautherie M, Albert E, editors. Biomedical Thermology, Progress in Clinical Biological Research. New York:
Alan R. Liss; 1982. p 687696.
2. Law HT, Pettigrew RT. Heat transfer in whole body
hyperthermia. Ann NY Acad Sci 1980;335:298310.
3. Robins HI, et al. A non-toxic system for 41.8 8C whole body
hyperthermia: results of a phase I study using a radiant heat
device. Cancer Res 1985;45:39373944.
4. Parks LC, Smith GV. Systemic hyperthermia by extracorporeal induction techniques and results. In: Storm FK, editor.
20.
21.
22.
Hyperthermia in Cancer Therapy. Boston: Hall; 1983.

pp 407446.
van der Zee J, van Rhoon GC, Faithfull NS, van den Berg AP.
Clinical hyperthermic practice: whole body hyperthermia. In:
Field SB, Hand JW, editors. An Introduction to the Practical
Aspects of Clinical Hyperthermia. London: Taylor & Francis;
1990.
Wust P, et al. Feasibility and analysis of thermal parameters
for the whole-body-hyperthermia system IRATHERM-2000.
Int J Hyperthermia 2000 JulAug; 16(4):325339.
Blazickova S, Rovensky J, Koska J, Vigas M. Effect of
hyperthermic water bath on parameters of cellular immunity.
Int J Clin Pharmacol Res 2000;20(12):4146.
Vaupel P, Kallinowski F. Physiological effects of hyperthermia. In:
Streffer C, editor. Hyperthermia and the Therapy of Malignant
Tumors. Berlin/Heidelberg, Germany: Springer-Verlag; 1987.
Dudar TE, Jain RK. Differential response of normal and tumor
microcirculation to hyperthermia. Cancer Res 1984;44(2):605
612.
Kong G, Braun RD, Dewhirst MW. Characterization of the
effect of hyperthermia on nanoparticle extravasation from
tumor vasculature. Cancer Res 2001;61:30273032.
Jain RK. Normalization of tumor vasculature: an emerging concept in antiangiogenic therapy. Science 2005;307(5706): 5862.
Urano M, et al. Effect of whole body hyperthermia on cell
survival, metastasis frequency, and host immunity in moderately and weakly immunogenic murine tumors. Cancer Res
1983;43:10391043.
Shen RN, et al. Whole-body hyperthermia decreases lung
metastases in lung tumor-bearing mice, possibly via a
mechanism involving natural killer cells. J Clin Immunol
1987;7(3):246253.
Roigas J, Wallen ES, Loening SA, Moseley PL. Heat shock
protein (HSP72) surface expression enhances the lysis of a
human renal cell carcinoma by IL-2 stimulated NK cells. Adv
Exp Med Biol 1998;451:225229.
Burd R, et al. Tumor cell apoptosis, lymphocyte recruitment
and tumor vascular changes are induced by low temperature,
long duration (fever-like) whole body hyperthermia. J Cell
Physiol 1998;177(1):137147.
Ostberg JR, Gellin C, Patel R, Repasky EA. Regulatory potential of fever-range whole body hyperthermia on Langerhans
cells and lymphocytes in an antigen-dependent cellular
immune response. J Immunol 2001;167(5):26662670.
Kraybill WG, et al. A phase I study of low temperature, long
duration (fever-range) whole body hyperthermia (WBH) in
patients with advanced solid tumors: correlation with mouse
models. Int J Hyperthermia 2002;18:253266.
Atanackovic D, et al. 41.8 degrees C whole body hyperthermia
as an adjunct to chemotherapy induces prolonged T cell activation in patients with various malignant diseases. Cancer
Immunol Immunother Epub 2002 Oct 18 2002;51(1112):603
613.
Barni S, et al. Lysosomal exocytosis induced by hyperthermia:
a new model of cancer death. III. effect on liver metastasis.
Biomed Pharmacother 1996;50:7984.
Hou K. The level changes of peripheral blood PGE2 and clinical
significance in patients with malignant tumor before and after
whole body hyperthermia. Proceedings of the 26th Congress of
the International Clinical Hyperthermia Society, September
912, 2004, Shenzhen, China, 2004; p 66.
Manjili MH, et al. Subjeck, Cancer immunotherapy: stress
proteins and hyperthermia. Int J Hyperthermia 2002;18(6):
506520.
Ialenti A, et al. Inhibition of cyclooxygenase-2 gene expression
by the heat shock response in J774 murine macrophages. Eur
J Pharmacol 2005;509(23):8996.
23. Repasky EA, Issels R. Physiological consequences of
hyperthermia: Heat, heat shock proteins and the immune
response. Int J Hyperthermia 2002;18:486489.
24. Ostberg JR, Repasky EA. Emerging evidence indicates that
physiologically relevant thermal stress regulates dendritic cell
function. Cancer Immunol Immunother [Epub ahead of print];
Apr 28, 2005.
25. Berry JM, Michalsen A, Nagle V, Bull JM. The use of esmolol
in whole-body hyperthermia: cardiovascular effects. Int J
Hyperthermia 1997;13(3):261268.
26. Robins HI, et al. Adjunctive therapy (whole body hyperthermia versus Ionidamine) to total body irradiation for the treatment of favorable B-cell neoplasms: a report of two pilot
clinical trials and laboratory investigations. Int J Radiat Onco
Biophys 1990;18:909920.
27. Ohno S, et al. Haematological toxicity of carboplatin and
cisplatin combined with whole body hyperthermia in rats.
Br J Cancer 1993;68:469474.
28. Haveman J, et al. Effects of hyperthermia on the peripheral
nervous system: a review. Int J Hyperthermia 2004;20(4):371391.
29. Calvert AH, et al. Early clinical studies with cis-diammine-1,1cyclobutane dicarboylate platinum II. Cancer Chemother
Pharmacol 1982;9:140147.
30. Sapareto SA, Dewey WC. Thermal dose determination in cancer
therapy. Int J Radiat Oncol Biol Phys 1984;10(6):787800.
31. Thrall DE, et al. Using units of CEM 43 degrees C T90, local
hyperthermia thermal dose can be delivered as prescribed. Int
J Hyperthermia 2000;16(5):415428.
32. Webb AG. Temperature measurement using nuclear magnetic
resonance. Ann Reports NMR Spectrosc 2001;45:167.
33. Stafford RJ, Hazle JD, Glover GH. Monitoring of high-intensity focused ultrasound-induced temperature changes in vitro
using an interleaved spiral acquisition. Magn Reson Med
2000;43(6):909912.
34. Engelhardt R. Summary of recent clinical experience in wholebody hyperthermia combined with chemotherapy. Recent
Results Cancer Res 1988;107:200224.
35. Lange J, Zanker KS, Siewert JR, Eisler K, Landauer B, Kolb
E, Blumel G. and Remy, W. Extracorporeally induced wholebody hyperthermia in conventionally incurable malignant
tumor patients Med Wochenschr. 1983;108(13):504509.
36. Robins HI, et al. Phase I clinical trial of melphalan and 41.8
degrees C whole-body hyperthermia in cancer patients. J Clin
Oncol 1997;15:158164.
37. Issels RD, Wilmanns W, editors. Recent Results in Cancer
Research, Vol. 107: Application of Hyperthermia in the
Treatment of Cancer. Berlin/Heidelberg: Springer Verlag; 1988.
38. Westermann AM, et al. Systemic Hyperthermia Oncologic
Working Group, A Systemic Hyperthermia Oncologic Working
Group trial, Ifosfamide, carboplatin, and etoposide combined
with 41.8 degrees C whole-body hyperthermia for metastatic
soft tissue sarcoma. Oncology 2003;64(4):312321.
39. Bull JM, et al. Chemotherapy resistant sarcoma treated with
whole body hyperthermia (WBH) combined with 1-3-bis(2chloroethyl)-1-nitrosourea (BCNU). Int J Hyperthermia 1992;
8(3):297304.
40. Herman TS, Sweets CC, White DM, Gerner EW. Effect
of heating on lethality due to hyperthermia and selected
chemotherapeutic drugs. J Natl Cancer Inst 1982;68(3):487
491.
41. Engelhardt R, et al. Treatment of disseminated malignant
melanoma with cisplatin in combination with whole-body
hyperthermia and doxorubicin. Int J Hyperthermia 1990;6(3):
511515.
42. Robins HI, et al. Phase I clinical trial of carboplatin and 41.8
degrees C whole-body hyperthermia in cancer patients. J Clin
Oncol 1993;11:17871794.
61
43. Douwes F, et al. Whole-body hyperthermia in combination

with platinum-containing drugs in patients with recurrent
ovarian cancer. Int J Clin Oncol 2004;9(2):8591.
44. Hegewisch-Becker S, et al. Whole-body hyperthermia
(41.8 8C) combined with bimonthly oxaliplatin, high-dose leucovorin and 5-fluorouracil 48-hour continuous infusion in
pretreated metastatic colorectal cancer: a phase II study.
Ann Onc 2002;13(8):11971204.
45. Hegewisch-Becker S, et al. Whole body hyperthermia (WBH,
41.8 8C) combined with carboplatin and etoposide in advanced biliary
tract cancer. Proc Am Soc Clin Oncol 2003; (abstr. 1247), 22:311.
46. Richel O, et al. Phase II study of carboplatin and whole body
hyperthermia (WBH) in recurrent and metastatic cervical
cancer. Gynecol Oncol 2004;95(3):680685.
47. Strobl B, et al. Whole body hyperthermia combined with carboplatin/paclitaxel in patients with ovarian carcinomaPhase-IIstudy. J Clin Oncol -Proc Am Soc Clin Oncol 2004;22(14S).
48. Elias D, et al. Heated intra-operative intraperitoneal oxaliplatin plus irinotecan after complete resection of peritoneal
carcinomatosis: pharmacokinetics, tissue distribution and tolerance. Ann Oncol 2004;15:15581565.
49. Bull JM, et al. Phase I study of long-duration, low-temperature whole-body hyperthermia (LL-WBH) with liposomal doxorubicin (Doxil), 5-fluorouracil (5-FU), & interferon-a (IFN-a),
Proc Amer. Soc Clin Oncol 2002;(Abst. 2126).
50. Larkin JM, A clinical investigation of total-body hyperthermia
as cancer therapy. Cancer Res, 1979;39(6 Pt 2):22522254.
51. Bull JM, et al. Update of a phase I clinical trial using feverrange whole-body hyperthermia (FR-WBH) cisplatin (CIS)
gemcitabine (GEM) metronomic, low-dose interferon-alpha
(IFN-alpha), Proceedings of the International Congress on
Hyperthermic Oncology, 20.-24.04., St. Louis, (Session 21
and Poster 689); 2004.
52. Barlogie B, Corry PM, Lip E, Lippman L, Johnston DA,
Tenczynski TF, Reilly E, Lawson R, Dosik G, Rigor B,
Hankenson R, Freireich EJ Total-body hyperthermia with
and without chemotherapy for advanced human neoplasms.
Cancer Res 1979;39(5):14811489.
53. Robins HI, et al. Phase I trial of human lymphoblastoid
interferon with whole body hyperthermia in advanced cancer.
Cancer Res 1989;49(6):16091615.
54. Gerad H, van Echo DA, Whitacre M, Ashman M, Helrich M, Foy
J, Ostrow S, Wiernik PH, Aisner J. Doxorubicin, cyclophosphamide, and whole body hyperthermia for treatment of advanced
soft tissue sarcoma. Cancer. 1984 Jun 15;53(12):258591.
55. Baba H, et al. Increased therapeutic gain of combined cisdiamminedichloroplatinum (II) and whole body hyperthermia
therapy by optimal heat/drug scheduling. Cancer Res
1989;49(24 Pt. 1):70417044.
56. Katschinski DM, et al. Optimization of chemotherapy administration for clinical 41.8 degrees C whole body hyperthermia.
Cancer Lett 1997;115(2):195199.
57. Hegewisch-Becker S, et al. Effects of whole body hyperthermia
(41.8 degrees C) on the frequency of tumor cells in the peripheral blood of patients with advanced malignancies. Clin
Cancer Res 2003;9(6):20792084.
58. Teicher BA, Holden SA, Ara G, Menon K. Whole-body
hyperthermia and lonidamine as adjuvant therapy to treatment with cisplatin with or without local radiation in mouse
bearing the Lewis lung carcinoma. Int J Hyperthermia
1995;11(5):637645.
59. Sumiyoshi K, Strebel FR, Rowe RW, Bull JM. The effect
of whole-body hyperthermia combined with metronomic chemotherapy on rat mammary adenocarcinoma metastases. Int
J Hyperthermia 2003;19(2):103118.
60. Overgaard J, et al. Randomised trial of hyperthermia
as adjuvant to radiotherapy for recurrent or metastatic
62
61.
62.
63.
64.
65.
66.
67.
68.
malignant melanoma, European Society for Hyperthermic
Oncology. Lancet 1995;345(8949):540543.
Hehr T, Wust P, Bamberg M, Budach W. Current and potential role of thermoradiotherapy for solid tumours. Onkologie
2003;26(3):295302.
Hildebrandt B, et al. Current status of radiant whole-body
hyperthermia at temperatures > 41.5 degrees C and practical
guidelines for the treatment of adults. The German Interdisciplinary Working Group on Hyperthermia. Int J Hyperthermia 2005;21(2):169183.
Hildebrandt B, et al. Whole-body hyperthermia in the scope of
von Ardennes systemic cancer multistep therapy (sCMT)
combined with chemotherapy in patients with metastatic
colorectal cancer: a phase I/II study. Int J Hyperthermia
2004;20(3):317333.
Robinson V, et al. Thermotherapy for treating rheumatoid
arthritis. Cochrane Database Syst Rev 1: CD002826; 2002.
van Soest H, van Hattum J. New treatment options for chronic
hepatitis C. Adv Exp Med Biol 2003;531:219226.
Pontiggia P, Rotella GB, Sabato A, Curto FD. Therapeutic
hyperthermia in cancer and AIDS: an updated survey. J
Environ Pathol Toxicol Oncol 1996;15(24):289297.
Li CY, Dewhirst MW. Hyperthermia-regulated immunogene
therapy. Int J Hyperthermia 2002;18(6):586596.
Okita A, et al. Efficiency of lipofection combined with hyperthermia in Lewis lung carcinoma cells and a rodent pleural dissemination model of lung carcinoma. Oncol Rep 2004;11:13131318.
Further Reading
Bakhshandeh A, et al. Year 2000 guidelines for clinical practice of
whole body hyperthermia combined with cytotoxic drugs from
the University of Lubeck and the University of Wisconsin. J
Oncol Pharm Practice 1999;5(3):131134.
Field SB, Hand JW, editors. An Introduction to the Practical Aspects
of Clinical Hyperthermia. London: Taylor & Francis; 1990.
Gautherie M, editor. Methods of External Hyperthermic Heating.
Berlin/Heidelberg: Springer Verlag; 1990.
Gautherie M, editor. Whole Body Hyperthermia: Biological and
Clinical Aspects. Berlin/Heidelberg: Springer Verlag; 1992.
Hahn GM. Hyperthermia and Cancer. New York: Plenum Press, 1982.
Hildebrandt B, et al. Current status of radiant whole-body
hyperthermia at temperatures > 41.5 degrees C and practical
guidelines for the treatment of adults. The German Interdisciplinary Working Group on Hyperthermia. Int J Hyperthermia
2005;21(2):169183.
Issels RD, Wilmanns W, editors. Recent Results in Cancer
Research, Vol. 107: Application of Hyperthermia in the
Treatment of Cancer. Berlin/Heidelberg: Springer Verlag;
1988.
Nussbaum GH, editor. Physical Aspects of Hyperthermia. American
Association of Physicists in Medicine Medical Physics Monograph
No. 8. New York: American Institute of Physics; 1982.
Guan J, et al. The clinical study of whole-body hyperthermia
(WBH) improving the survival state of patients with advanced
cancer. Proc 26th Congress of the International Clinical
Hyperthermia Society, Sept. 912, 2004, Shenzhen, China;
2004; p 66.
Kurpeshev OK, Tsyb AF, Mardynsky YS. Whole-body hyperthermia
for treatment of patients with disseminated tumors- Phase II. In:
P.H. Rehak, K.H. Tscheliessnigg, editors. Proceedings 22nd.
Annual Meeting of the European Society for Hyperthermic
Oncology, June 811, 2005, Graz, Austria, 2005; p 103.
Hou K. Assessment of the effects and clinical safety of the treatment of advanced malignant tumor with extracorporeal whole
body hyperthermia. Proceedings of the 26th Congress of the
International Clinical Hyperthermia Society, Sept. 912, 2004,

Shenzhen, China; 2004 p 71.
See also BIOHEAT TRANSFER; HEAT
AND COLD, THERAPEUTIC; HYPERTHER-
MIA, INTERSTITIAL; HYPERTHERMIA, ULTRASONIC; RADIATION DOSIMETRY FOR

ONCOLOGY.
DIMPI PATEL
DHANUNJAYA LAKKIREDDY
ANDREA NATALE
The Cleveland Clinic Foundation
Cleveland, Ohio
INTRODUCTION
The use of elevated temperature as a form of medical
treatment has been fairly ubiquitous across cultures
throughout the course of time. The earliest record of heat
for therapeutic use was found in an Egyptian surgical
papyrus dated to 3000 BC (1). Hippocrates, considered
by many to be the father of medicine, used heat to treat
breast cancer. He based his practice of medicine on an
ancient Greek ideology that advises using heat after trials
of surgery and medications have failed (2). German physicians in the 1800s noted cases where cancer patients had
developed high fevers secondary to infections that resulted
in a miraculous disappearance of their tumors (3). These
observations provided inspiration for the development of
several techniques that attempted to induce hyperthermia.
One such popular method entailed wrapping a patients
body in plastic and then dipping him in hot wax. Another
popular technique involved removing a portion of the
patients blood, heating it, and then transfusing the warmed
blood back to the patients body, thereby creating systemic
hyperthermia (4). These treatments had varied success
rates, often culminating in fatality, and were subsequently
discarded. Thus, the interest in hyperthermia lessened in
the face of more conventional cancer treatments (e.g., chemotherapy and radiation). The current revival of interest in
hyperthermia has resulted from a combination of clinicians
searching for a therapeutic mode other than chemotherapy
and radiation, in tandem with several preliminary randomized clinical trials in a small selected group of patients that
have shown marked improvement in disease states with the
use of either hyperthermia alone or particularly as an
adjuvant to other more traditional modalities.
Traditionally, conventional hyperthermia has been
defined as a therapeutic elevation of whole body temperature
or target tissue while maintaining low enough temperatures
to avoid tissue coagulation (3). This definition of hyperthermia can be broadened to include the therapeutic elevation of
temperature to cause tissue destruction and coagulation,
such as that implemented in HIFU (high intensity focus
ultrasound) procedures. Classically, microwaves, radio frequency (RF), electromagnetic radiations, or ultrasounds have
been used to heat tissue to 4044 8C (5). This article compares
and contrasts electromagnetic waves to ultrasonic waves as a
heating modality, explain the physics behind ultrasound
generation, and explores the thermal and mechanical biophysics involved with ultrasound delivery to tissue. Then, the
medical fields that are currently benefitting from conventional ultrasound hyperthermia and HIFU are considered,
and finally some of the many applicators involved with
thermal ultrasound delivery are evaluated.
ULTRASOUND VERSUS ELECTROMAGNETIC RADIATION
Electromagnetic waves were often used in various applications of conventional hyperthermia treatments. However,
ultrasound has emerged as a better option because of its
shorter wavelength and lower energy absorption rate,
which make it easier to control and to localize the area
that is being heated. For example, for a half-power penetration depth of 4 cm, the ultrasound wavelength in tissues
(e.g., muscle) is 1 mm; however, electromagnetic wavelength required for the same transmission is 500 mm.
Focusing energy into a volume smaller than a wavelength
is generally not possible. Using 500 mm (40 MHz) of
electromagnetic waves to heat a tumor that is situated
4 cm below the skin with proportions of 6 cm in diameter in
the upper abdomen results in a temperature elevation of
the entire abdomen including the spleen, liver, and all
major vessels. More than one-half of the bodys cardiac
output circulates through the abdominal area, and this
widespread heating results in a systemic elevation of temperature, thereby limiting the use of electromagnetic
radiation for tumors in the body cavity (3). Electromagnetic
waves are currently limited to regional hyperthermia and
treating very superficial tumors (6). Conversely, ultrasound that has a wavelength of 1 mm can be focused within
the area of the tumor, thus allowing less energy to be
radiated to other areas of the body, resulting in less
damage to surrounding healthy tissue. The current fabrication technology allows for practical applicator dimensions and multiple transducer configurations that makes
it possible to control and shape a wide variety of ultrasound
beams. The use of focused transducers or electronically
phased arrays allow for better localization and temperature control of the target tissue (7). In contrast to these
positive attributes, high acoustic absorption at bonesoft
tissue interface and reflection from gas surfaces may make
certain therapeutic scenarios difficult.
GENERATION AND PROPAGATION OF ULTRASOUND
Ultrasonic Transducers
In order to generate ultrasonic waves for tissue warming, a
transducer containing piezoelectric crystals is required.
Piezoelectric crystals are found in Nature or can be artificially grown. Quartz and synthetic ferroelectric ceramics
(e.g., lead metanobiate, lead zirconate, and titanates of
barium) all have strong piezoelectric properties (8). The
ceramic most commonly used in the fabrication of ultrasound transducers is synthetic plumbium zirconium
titante (PZT). Transducers are manufactured by applying
an external voltage to these ferroelectric materials to
orient their internal dipole structure. They are then cooled
to permanently maintain their dipole orientation. Finally,
63
they are cut into any desired shape, such as spherical bowls
for focused ultrasonic fields (3,8). Ultrasound transducers
have electrodes attached to the front and back for application and detection of electrical fields. With the application
of an alternating electrical field parallel to the surface of
piezoelectric material, the crystals will contract and
vibrate for a short time with their resonant frequency.
The frequency at which the transducer is able to vibrate
is indirectly proportional to its thickness; higher frequencies are a result of thinner transducers, lower frequencies a
result of thicker transducers (8).
Piezoelectric crystals are able to contract or expand
when an electrical field is applied to them because dipoles
within the crystal lattice will realign themselves as a result
of attractive and repulsive forces causing a change in
physical dimension of the material in the order of nanometers (electrostriction or reverse piezoelectric effect).
When echos are received, the ultrasound waves will compress and expand the crystals (8). This mechanical stress
causes the dipoles to realign on the crystal surface creating
a net charge (piezoelectric effect) (Fig. 1).
Transducers function optimally when there is broad
bandwidth in the frequency domain and short impulse
response in the time domain. Also, when there is little
electroacoustical conversion inefficiency, and little mismatch between the electrical impedances of the generator
and the transducer (3,9). A transducers ability to transmit
energy is dependent on the characteristics of acoustic impedances and its contact medium. Both the density of the
material and propagation velocity of ultrasound waves will
determine its impedance. When both impedances match,
then less energy is lost through reflection back into the
transducer. For example, at the interface between air and
the transducer, most of the energy will be reflected back to
the transducer and will travel to the opposite direction
because air has 16 times less impedance than the transducer. If the transducer is a half wavelength in thickness,
the reflected wave arrives at the opposite surface in phase
with the direction of its motion and can then be transmitted
into the medium. Since the efficiency at which a transducer
transmits energy has a direct relationship to the degree of
impedance match, efficiency can be increased significantly
by adding an impedance matching layer of a quarter wavelength thickness, subsequently making the characteristic
impedance equal to the geometric average of those of the
transducer and the loading medium (3,8).
RADIATION FIELD OF ULTRASONIC TRANSDUCERS
The radiation field of an ultrasonic transducer depends on its
physical properties and the transmission characteristics of
the medium through which it will pass. Conventional planar
transducers create a nonfocused field, whereas some modifications to the same transducer can create a focused field.
NONFOCUSED FIELDS
Planar Transducers
Ultrasonic waves that are radiated from the transducer
surface can be described as a tightly packed array of
64
Figure 2. A planar transducer operating in a continuous wave

mode. (a) The envelope containing almost all of the ultrasonic
energy. (b) The relative intensity of the ultrasonic beam along a
central axis. The intensity across the field fluctuates greatly at
small distances from the surface of the transducer. At greater
distances along the central axis the intensity distribution across
the field stabilizes and deteriorates with beam divergence. (c) Ring
diagrams illustrating the energy distribution at positions
indicated in (b). In the near field, the ring beam pattern is
collimated, but at greater distances from the transducer surface
the beam diverges. (Published with permission from Ref. 3).
Figure 1. A piezoelectric compound consists of aligned molecular

dipoles. (a) At equilibrium, the dipoles are arranged in a
configuration that results in a neutral net charge. When the
piezoelectric compound is mechanically stressed (e.g., an
ultrasound wave) the element changes its physical dimensions. At
peak pressure amplitudes, the element will contract. When no stress
is placed upon the element it is in equilibrium. At peak rarefaction,
the element will expand. This realignment of dipoles results in the
production of a net positive or negative surface charge. (b) When an
electrical field is applied to the piezoelectric element the dipoles can
be realigned in response to attractive or repulsion forces. This
rearrangement results in either expansion or contraction of the
element. In the absence of an applied electrical field the element
is in equilibrium and has a neutral net charge. (Published with the
permission from Ref. 8).
separate point sources of sound energy (Fig. 2a). Each of

these points emits a spherical wavelet (Fig. 3). These
waves interact both constructively and destructively
creating a diffraction pattern. Any point in the medium
is a compilation of all the sources that reach that target at
that period of time. This diffraction pattern can be calculated using Huygens principle. Two separate transducers
whose emission fields interact in the same media are
subject to the same laws of construction and destruction.
Planar transducers operating in continuous wave mode
are shown in (Fig. 2). In the Fresnel zone or the near field,
the beam energy distribution is collimated, which is a
result of the many destructive and constructive interactions of the spherical wavelets (Figs. 2c and 3). The
beam path is a function of the dimension of the active

part of the transducer surface, thus the beam diameter
that is converging at the end of the near field is approximately one-half of the size of transducer diameter.
The intensity and pressure amplitudes fluctuate greatly
at small distances from the surface transducer (Fig. 2b).
As the distance from the transducer surface increases,
the beam path diverges (Fig. 2c and 3). In large diameter,
high frequency transducers, there is less beam divergence in the far field. After a certain distance from the
transducer surface, the intensity stabilizes; however,
intensity along the axis deteriorates along with beam
divergence (Fig. 2b). Circular, square, and rectangular
transducers have similar fields; albeit, circular transducers have more pronounced fluctuations of intensity
in the near field (3,8).
Constructive and destructive interference

patterns confine beam in the near field
Direction of beam
Near field
Far field
Figure 3. Ultrasonic waves that are radiated from the transducer surface are described as a tightly packed array of separate
point sources. Each of these points radiates a spherical wavelet
(Huygens principle). These spherical wavelets will interact constructively and destructively. In the near field, these interactions
result in a convergent beam pattern. In the far field, the beam pattern
diverges. (Published with permission from Ref. 8).
FOCUSED FIELDS
65
2.7 MHz
Single-Element Transducers
100%
85%
Vertical (mm)
When an ultrasonic wave travels through different media,

the laws of geometric optics can be applied. Ultrasonic
waves can be reflected, refracted, and scattered. When
there is a high degree of impedance mismatch between
the generator and transducer, ultrasonic waves will be
reflected back into the transducer. The angle of reflection
is equal to the angle of impedance, much like that of a
mirror. Single element transducers can be focused by using
a curved acoustic lens or a curved piezoelectric element (8).
When an ultrasonic wave goes through two media with
different propagation velocities there is a certain degree of
refraction. Ultrasonic propagation through water is
1500 ms1. In order to focus the ultrasound field, a lens
of plastic (e.g., polystyrene), which has a higher propagation
velocity, is placed between the transducer and the water
media; these converging lenses are then concave to the
water media and at plane with the transducer interface.
In an unfocused transducer, the focal length is directly
proportional to the transducer frequency and diameter. In
a focused single element transducer, the focal distance is
brought closer to the transducer surface. The focal distance
is defined as the distance between the transducer surface
and the portion of the beam that is narrowest. The focal
zone, which is the area of best lateral resolution, is defined
as the area at which the width of the beam is less than two
times the width at the focal distance (3,8) (Fig. 4). The
focal zone is dependent on the aperture and the wavelength of the ultrasound. The focal area through which
84% of the ultrasound passes is two to four wavelengths in
hyperthermia systems. With ultrasonic transducers, the
intensity distribution dimensions are a function of frequency and aperture. Therefore, the larger the aperture,
the shorter the focal region, the higher the frequency and
the smaller the diameter of the beam (Fig. 5). Ceramic
curved bowl-shaped transducers, while more efficient
than a lens, do not have the versatility of a lens. Once
a ceramic bowl is fabricated, the focal length is set. Lenses
can be interchanged creating a variety of focal lengths
with one transducer (8).
70%
50%
20%
Lateral (mm)
Figure 5. The intensity distribution in the focus of an ultrasound
transducer. The diameter and the length are a function of
frequency. The lower the frequency, the larger the diameter,
and the smaller the aperture, the longer the focal region.
(Published with permission from Ref. 3).
Multielement Transducers
Curved element
Focal zone
Figure 4. The focal zone is the area of optimal lateral resolution.

The use of a curved element or an acoustic lens allows the focal
distance to be brought closer to the transducer surface. The use of
a curved element decreases the beam diameter at the focal
distance and increases the angle of beam divergence far field.
The focal zone, which is the are of best lateral resolution, is
defined as the area at which the width of the beam is less than
two times the width at the focal distance. (Published with
Linear array transducers contain 256512 narrow piezoelectric elements. They produce a beam by firing a portion
of the total number of elements as a group. If a single
element were fired the beam pattern would be divergent in
the near field. By firing a group of elements, it is possible to
focus and converge the beam. All the individual beams
interact both constructively and destructively to produce a
collimated beam. A phase array transducer is composed of
64128 elements. The ultrasound beam can be steered and
focused without moving the transducer by electrically
activating the separate elements on the transducer surface
at slightly different times (8).
66
Table 1. Acoustic Impedancea
Tissue
Air
Lung
Fat
water
Kidney
Blood
Liver
Muscle
Skull bone
Z, rayls
0.0004
0.18
1.34
1.48
1.63
1.65
1.65
1.71
7.8

106
106
106
106
106
106
106
106
106
Measured in rayls. z pc (z impedance,

p sound pressure, c speed of sound), for
air, water, and selected tissues. (Published
with permission from Ref. (8).
Table 2. Density and Speed of Sound in Tissues and

Materials for Medical Ultrasound.a
Material
Air
Lung
Fat
water
Soft tissue
Kidney
Blood
Liver
Muscle
Skull bone
PZTb
The journey of an ultrasound wave through human tissue

is sometimes arduous. As the waves propagate through
different biological media, they are subject to reflection,
refraction, scattering, and absorption (8). When there is a
difference in acoustic impedance between the boundaries of
two tissues, reflection occurs (Table 1). There is 100%
reflection at the airskin interface. However, if a coupling
medium (e.g., gel) is used, reflection is reduced to 0.1%.
When the beam is not perpendicular to tissue boundary,
the transmitted ultrasound energy undergoes a change in
direction at the boundary; this directional change is termed
refraction. As waves propagate through tissue they must
overcome internal friction resulting in a loss of energy. The
mechanical energy that is lost is converted to heat, which is
termed absorption. At higher frequencies, ultrasonic waves
move quicker, thus forcing the molecules to move against
each other creating friction. The more these molecules
move, the more energy is consumed or absorbed, and
subsequently will be converted to heat. The speed at which
the ultrasonic wave travels is dependent on the arrangement of the molecules. If they are densely arranged, they
will collide sooner when they are exposed to a stimulus, and
will lose energy quickly and at shorter distances. Ultrasonic waves can travel through the skin without much
absorption until they reach tissues with high collagen
content, (e.g., bone, periosteum, ligaments, capsules, fascia, and tendons). Ultrasonic waves travel through most
solid tissues at a speed of 15001600 ms1. Its velocity in
fat and postmenopausal breast tissue may be as low as
1450 ms1 and the lens of the eye 1600 ms1. As a
general rule, ultrasonic waves move through soft tissue
with relatively little reflection or refraction (3,8) (Table 2).
Ultrasonic speed through bone is 4080 ms1. Bone
easily absorbs ultrasonic energy and reflects it to tissues
that are located at the bonetissue interface. Since bone
absorbs ultrasonic energy so readily, it heats up very
quickly, consequently making it harder to control temperature. Thus, bone and its immediate surrounding tissue
were once considered problematic for therapeutic use of
ultrasonic hyperthermia (3,8). Nevertheless, in a recent
study, Moros et al. noted that the presence of underlying
c, ms1
c, mms1
1.2
300
924
1000
1050
1041
1058
1061
1068
1912
7500
330
600
1450
1480
1540
1565
1560
1555
1600
4080
4000
0.33
0.60
1.45
1.48
1.54
1.57
1.56
1.55
1.60
4.08
4.00
a
b
PROPAGATION OF ULTRASOUND IN BIOLOGICAL

TISSUES
Density, kgm3
Published with permission from Ref. 8.

PZT, lead cisconate .inanate
bone in superficial unfocused ultrasound hyperthermia

could actually be exploited to induce more uniform and
enhanced temperature distributions in superficial target
volumes. In particular, they have shown that the presence
of bone in superficial target volumes enhances temperature
elevation not only by additional direct power deposition
from acoustic reflection, but also from thermal diffusion
from the underlying bone (10).
The intensity at which an ultrasonic beam is transmitted has an effect on target tissue temperature. Intensity
is defined as the amount of power per unit area. Doubling
the amount of power used will result in quadrupling the
intensity. Ultrasonic waves will lose intensity as they
propagate further into the tissue. The attenuation coefficient is the relative intensity loss per centimeter of travel in
a given medium (Table 3). Beam divergence, absorption,
and scattering will also cause a loss in intensity of the
propagating beam. The absorption coefficient of the tissue
being exposed to us determines the target temperature
that tissue will reach. The absorption coefficient is dependent on the density of the tissue and will linearly increase
at higher frequencies. The absorption coefficient in soft
tissue is 410 times lower than that of bone, and therefore
bone heats more quickly (3). At short exposure times (e.g.,
< 0.1 s), temperature and intensity are directly propor-
Table 3. Attenuation Coefficients for Selected Tissues at

1 MHz.a
Tissue
Composition
Water
Blood
Soft tissues
Brain
Liver
Fat
Smooth muscle
Tendon
Bone, cortical
Lung
a
Attenuation Coefficient
1 MHz beam, dBcm1
0.0002
0.18
0.30.8
0.30.5
0.40.7
0.51.8
0.20.6
0.91.1
1326
40
tional. However, as the time intervals increase, other

factors in addition to intensity, (e.g., blood perfusion), must
be considered. An approximate estimate of the ultrasonic
energy requirements for heating a target volume to therapeutic temperature depends on assessing thermophysical
properties of that tissue, intensity of the ultrasound beam,
ultrasonic absorption coefficient, and additional factors
(e.g., blood circulation to target tissue) (3,8,11). The thermal index is defined as the ratio of acoustic power created
by the transducer to raise the target area by 1 8C. This is
calculated by using an algorithm that takes into account
the ultrasonic frequency, beam area, and the acoustic
power output of the transducer (8).
Ultrasonic waves act on tissues thermally and mechanically. Mechanical effects on tissues via ultrasound include
acoustic torque, acoustic streaming, radiation force, stable
cavitation, and unstable cavitation (11). Any object situated within an acoustic field will be subject to acoustic
pressure and acoustic force. Acoustic pressure in a standing wave field is inversely proportional to velocity. Acoustic
torque results from variations in the acoustic field, which
can be described as a time-independent twisting action.
Acoustic torque causes a rotational movement of cells and
intracellular organelles in the medium. Acoustic streaming
describes the movement of fluid in an ultrasonic field. The
compression phase of an ultrasonic wave deforms tissue
molecules. Radiation force affects gas bubbles that are in
the tissue fluids. Negative pressure induces the bubbles
originally dissolved in the medium to fall out of solution.
With positive and negative pressure wave fluctuations,
these bubbles expand and contract without reaching critical size (stable cavitation). Unstable cavitation occurs
when bubbles collapse violently under pressure after growing to critical size due to excessive energy accumulation.
This implosion produces large, brief local pressure and
temperature release, as well as causing the release of free
radicals. Organs that are air-filled, (e.g., the lungs or
intestines), are subject to greater chance of unstable cavitation. Unstable cavitation is somewhat random, and as
such it may lead to uncontrollable tissue destruction (8,11).
Bubble growth can be limited by low intensity, high frequency, and pulsed ultrasound. Higher frequency means
shorter cycle duration so time for bubble growth is regulated. Pulsed ultrasound restricts the number of successive
growth cycles and allows the bubble to regain its initial size
during the off period. The mechanical index estimates the
possibility of cavitation occurrence. The mechanical index
is directly proportional to peak rarefaction pressure, and
inversely proportional to the square root of the ultrasound
frequency (8).
67
supported by a plethora of studies that demonstrate that

heat on cell lines and on animal tumor transplant models
can result in tumor regression; however, it is rarely used
alone because its efficacy is greatly potentiated in combination with radiation or chemotherapy. Conventional
hyperthermia treatments elevate target tissue temperatures to 4246 8C (12). Treatment times are usually
between 30 and 60 min. Treatment applications are
administered once or twice a week and are applied in
conjunction with or not long after radiation. In all of the
recent phase III trials, a sequential delivery scheme was
used. This means that radiation and hyperthermia were
administered separately, with radiation preceding
hyperthermia treatments (13). Tumoricidal effects in
vivo are achieved at temperatures between 40 and
44 8C (5). Large tumors often have an inadequate blood
supply and resultantly, have difficulty meeting their
requirements for oxygen and nutrients. This situation
creates a hypoxic environment that is low in pH (23)
(3,5,14). When tumor cells are heated to therapeutic
temperatures, their cellular metabolic processes are
accelerated, thereby further increasing the demands
for oxygen and nutrients in an already depleted environment. Most tumor cells are unable to reproduce in this
hostile environment, resulting in termination of tumor
growth and shrinkage of the tumor (5,15). In temperatures > 40 8C, protein denaturation has been observed as
the main mechanism of cellular death. Widespread protein denaturation results in structural changes in the
cytoskeleton and the cell membrane, and in enzymes that
are necessary for deoxyribonucleic acid (DNA) synthesis,
cellular division, and cellular repair (5). Hyperthermic
efficacy is a function of temperature and exposure time.
To quantify, at temperatures > 42.5 43 8C, the exposure
time can be halved with each 1 8 temperature increase to
give an equivalent cell kill (5,16). Healthy tissues remain
undamaged at temperatures of 44 8C for a 1 h duration
(5,17). The exceptions are central nervous tissues, which
suffer irreversible damage after being exposed to heat at
temperatures ranging from 42 to 42.5 8C for >4060 min
(5,18). Peripheral nervous tissue that has been treated
for > 30 min at 44 8C or an equivalent dose results in
temporary functional loss that is reversed in 4 weeks
(5,19). Therefore, since a small difference in temperature
produces a large difference in the amount of cells killed,
it is important to be able to have good control on the site
and duration of heat delivery to reduce the damage to
surrounding healthy tissue.
RADIATION COUPLED WITH CONVENTIONAL

HYPERTHERMIA
MEDICAL APPLICATIONS OF CONVENTIONAL
HYPERTHERMIA
Ultrasound as a heating modality has been used in
several different medical fields. It is used in treating
sprains, bursitis, joint inflammation, cardiac ablations,
and in gynecology. However, the main area conventional
hyperthermia is currently used is in oncology. The use of
conventional hyperthermia as an oncologic treatment is
While hyperthermia independently has been found to have

antitumor effects, its efficacy is greatly potentiated when
coupled with radiation. Cells that are in a low pH hypoxic
environments, those that are in the S or M phases of cell
division, and those that are malnourished are relatively
resistant to radiation (5,7). Hyperthermia increases
radiation damage and prevents cellular repair of damaged
DNA (5,16). Hyperthermia increases blood perfusion via
68
vasodilation which results in increased oxygenation, thus

allowing increased radiosensitivity (5,7,16). Response
rates with hyperthermia alone are 15%, with radiotherapy 35%, with combined radiotherapy, and hyperthermia
70% (20). There have been many U.S. and European
clinical trials that support substantial improvement in
patients who have been treated with a combination of
radiation and hyperthermia. Examples of some recent
trials include randomized multiinstitutional phase III
trials for treating melanoma (20,21), glioblastoma multiforme (20,22), chest wall recurrence of breast cancer
(20,23), head and neck cancer (20,24,25), head and neck
in superficial measurable tumors (20,26,27), in various
recurrent persistent tumors (20,28), cervical cancer (29),
uterine cancer (30) and in locally advanced pelvic tumors
(20,31) (Table 4). Trial success rates were very dependent
on the uniformity of temperature delivery. In the past,
trials had often provided mediocre results because
temperatures were 11.5 8C too low and consequently
not able achieve adequate tumoricidal levels (7). It is often
difficult to uniformly heat larger tumor (3,7). When radiation and hyperthermia are used simultaneously excellent
radiation delivery is achieved, often resulting in tumor
regression; however, its delivery is equally as toxic to
healthy cells that necessitate the need for a very precise
delivery system.
CHEMOTHERAPY IN CONJUNCTION WITH

CONVENTIONAL HYPERTHERMIA
Chemotherapeutic efficacy is enhanced by hyperthermia
(5,20,34,35) (Table 5). As areas are heated, perfusion is
increased, thus allowing an increase in drug concentrations in areas of the tumor that are poorly vascularized,
increased intracellular drug uptake, and enhanced DNA
damage. Drugs (e.g., mitomycin C, nitrosureas, cisplatin,
doxorubicin, and mitoxantrone) are subject to less drug
resistance when used with heat. The synergistic effect of
chemotherapy and hyperthermia was demonstrated in
virtually all cell lines treated at temperatures >40 8C for
alkylating drugs, nitroureas, and platin analogs dependent
on exposure time and temperature. Chemotherapeutic
agents can be revved up 1.210 times with the addition
of heat (5). In vivo, experiments showed improvement
when doxorubicin and mitoxantrone were combined with
hyperthermia. However, antimetabolites vinblastine, vincristine, and etoposide did not show improvement with the
addition of hyperthermia. In animal studies, increased
toxicities were seen in skin (cyclophosphamide, bleomycin),
heart (doxorubicin), kidney (cisplatin, with a core temperature >41 8C), urinary tract (carmustine, with core temperatures >41 8C), and bone marrow (alkylating agents
and nitroureas) (5,34). Lethal toxicity was enhanced when
systemic hyperthermia was applied in combination with
cyclophosphamide, methyl-CCNU, and carmustine (5). The
success of hyperthermia and chemotherapy combinations
depends on the temperature increase in the organs for
which the drug is used and its subsequent toxicity, all of
which are can be influenced by the accuracy of the heating
device and the operator.
MODES OF CONVENTIONAL HYPERTHERMIA

APPLICATION
Externally Applied Techniques
In the past, single planar transducers were used to apply
local heat. A disk shaped piezoelectric transducer (range
from 0.36.0 MHz in frequency and up to 16 cm in diameter) is mounted above a chamber of cooled degassed
water. This device has a coupling chamber which allows
water to circulate (3) (Fig. 6). It is coupled to the body via a
plastic or latex membrane. Unfortunately, these types of
devices are unable to achieve homogenous therapeutic
thermal levels. The reason is that this system uses an
unfocused energy source. When an unfocused energy
source is applied to the skin, the intensity and the temperature will be the highest at the contact point and will
subsequently lose intensity as it travels deeper into the
tissue. However, by cooling the skin, the hot spot is
shifted to the subcutaneous fatty tissue that is poorly
vascularized. Fat is an insulator and as a result much
energy is conserved rather than lost. Furthermore, cooling
the skin will produce vasoconstriction which conserves
even more heat and facilitates the heating of deeper tissues
(3) (Figs. 7, 8a and b). However, even with this strategy
adequate temperatures could not be reached. The reason
for this is that large tumors often consist of three zones, a
central necrotic core, an intermediate zone that is normally
perfused, and a marginal zone that has a greater number of
vessels due to proliferation induced angiogenesis. Due to
the abundance of vasculature on the marginal surface,
much heat is dissipated to the surrounding tissue. The
relatively avascular center will heat to a higher temperature than the marginal or intermediate zone because there
is little dissipation of heat, creating hot spots (Fig. 9) (7,54).
Thus, it is not possible to therapeutically heat a tumor with
a single planar source. This theory is substantiated by a
significant number of clinical trials (7,5561). Most trials
reported that patients had some difficulty with doselimiting pain, extensive central tumor necrosis, blistering,
and ulceration (7). Conversely, a focused ultrasound source
localizes energy within the tumor volume while sparing the
surrounding tissue (Fig. 10). The use of a focused beam
allows for homogenous heating and higher intensity which
allows the generation of greater temperatures within the
target area. Attenuation and beam divergence cause rapid
deterioration of intensity beyond the focal zone (3) (Fig. 11
a and b). Focused ultrasound sources overcome some of the
limitations of planar heating. Focusing allows for controlling the amount of heat that is delivered to the poorly
perfused areas thus limiting hot spots and some of the side
effects. For a heating system to be successful clinically on a
large scale, it must account for geometric and dimensional
variations of target tissue, possess the ability to heat the
sites that need it, and avoid side effects and complications
as much as possible (3).
Technological advances in hyperthermia devices have
paved the way for better therapeutic options. The use of
mosaics or separately controlled transducers allowed better spatial and temperature control to target bulky irregularly shaped tumors. The multielement planar
69
Table 4. Hyperthermia and Radiation Clinical Trials

Reference/
name of
trial
(26,32)
RTOG
(25)
(24,33)
(21) ESHO-3
(23) MRC/
ESHO-5
(31)
(28)
(25)
(29)
Tumor Entity (stage)

Head and neck
(superficial measurable
tumor)
Head and neck untreated
locoregional tumor
Head and neck

(N3locoregional tumor)
Melanoma (skin
metastases or recurrent
skin lesions)
Breast cancer (local

recurrences or inoperable
primary lesions)
Rectal cancer
Type of Trial
Prospective
randomized
multicenter
Prospective
randomized
Prospective
randomized
Prospective
randomized
Multicenter
Randomized
multicenter
Prospective
randomized
multicenter
No. of
Patients
106
65
41
70
306
143
Bladder cancer
101
Cervical cancer
114
Various (recurrent or
progressive lesions)
Gioblastoma
(postoperative)
Stage IIIB
uterine cervix
(27)
Superficial tumors
(30)
Uterine cervical
Prospective
randomized
multicenter
Prospective
randomized
Prospective
randomized
Prospective
randomized
Prospective
randomized
multicenter
174
Type of
Hyperthermia
Superficial
(915 MHz
microwave)
Superficial
(27-12 MHz
microwave)
Superficial
(280 MHz
microwave)
Superficial
(various
techniques)
Superficial
(various
techniques)
Deep regional
HT (various
techniques)
Interstitial HT
(300-2450 MHz
microwave
or RF)
79
Interstitial HT
40
Deep regional
HT
Results of
Control Arm
(RT only)a
Results of
Significance
Hyperthermia
of Results
Arm (RTHT)a
( p<0.05)
34% CR
34% CR
32% DR
55% CR
19% DFS at
1.5 years
41% CR
33% DFS
at 1.5 years
83% CR
24% LRFS
0% OS at
5 years
35% CR
68% LRFS
53% OS at
5 years
62% CR
28% LRFS at
5 years
41% CR
46% LRFS at
5 years
59% CR
ca. 30% LRFS

ca. 40% AS
at 2 years
15% CR
ca. 50% LRFS

ca. 40% AS at
2 years
21% CR
22% OS at
3 years
51% CR
22% OS at
3 years
57% CR
27% OS at
3 years
54% CR
13% OS at
3 years
73% CR
28% OS at
3 years
83% CR
51% OS at
3 years
57% CR
34% OS at
2 years
15% OS at
2 years
50% CR
35% OS at
2 years
31% OS at
2 years
80% CR
79.7% CR at
3 years
66.1% CR
73.2% CR
122
EM
45% CR at
3 years
42.3% CR
110
RF
68.5% CR
AS actuarial survival; CR complete remission; DFS disease free survival; HT hyperthermia; LRFS local relapse free survival; OS overall
survival; RF radio frequency electric currents; RT radiotherapy. (Published with permission from Ref 20).
ultrasound applicators met these demands and are capable

of treating tumors at depths up to 8 cm. The multisector
applicator allows for heating to the edge of the aperture
and the acoustic beams are nondiverging in the near field,
thus allowing large tumor heating with lateral measurements of 15 15 cm. Each of these 16 sectors can be varied
from 0 to 100% power to uniformly heat across the tumor. If
an area of the tumor is too difficult to treat, more energy
70
Table 5. Hyperthermia and Chemotherapy Clinical Trials

Reference
(36)
(37)
(38)
Tumor Entity
Oesophagus
cancer
(preoperative)
Oesophagus
cancer
(preoperative)
Stomach cancer
Type of
Trial
(40,41)
Pancreatic cancer
Sarcomas
(pretreated with
chemotherapy)
Type of
Hyperthermiaa
Type of
Chemotherapya
Resultsa
Phase II
32
localHT/
Endoluminal MW
CDDP Bleo
Cyc
8 CR/13 PR
(65% RR)
Phase III
20
localHT/
Endoradiotherm
CDDP Bleo
1 CR/5 PR/4 MR
(50% RR);
FHR (41.2%)
0CR/5 PR/0 MR
(25% RR);
FHR (18.8%)
3 CR 10 PR
(39% RR)
3 CR 5 PR
(36% RR)
27..3% survival
at 1 year
6 pCR 4PR 4FHR
(37% RR)
20
Control
CDDP Bleo
33
RHT/thermotron
Mitomycin 5FU
22
8 MHz
Mitomycin 5FU
Phase II
77
RHT 13.5 MHz
Phase II
(RHT 86)
38
RHT/BSD 1000
60-110 MHz
Mitomycin 5FU /
- immunostimulation
VP16 IFO
Phase II
pancreatic cancer
(39)
No. of
Patients
(42,43)
High risk soft

tissue sarcomas
Phase II
(RHT 91)
59
RHT/BSD 2000
80-110 MHz
VP16 IFO ADR
(44)
High risk soft

tissue sarcoma
112
Soft tissue
sarcoma
Sarcoma/
teratomas
(metastatic)
Sarcoma
(metastatic)
Refractory
cancers
(advanced or
metastatic)
Pediatric
sarcomas
55
RHT/BSD 2000
80-110 MHz
(randomized)
ILP with HT
VP16 IFO ADR
(45)
Phase III
(EORTC
62961)
Phase II
9pCR 4PR 8FHR

(32% RR)
ICR/6pCR 8PR 13 MR
(47%)
OS: 46% at 5 years
(08/00)
TNF IFN L-PAM
10CR/35PR (82% RR)
Phase I/II
19
WBH
IFO CBDCA
6PR (32% RR)
Phase II
12
WBH
IFO CBDCA VP16
7PR (58% RR)
Phase I
16
WBH
(Aquatherm)
Phase II
34
RHT/BSD 2000
80-110 MHz
V16 IFO CBDCA
Phase II
10
RHT/BSD 2000
80-110 MHz
CDDP VP16 IFO

(PEI)
Phase II
23
CDDP (weekly)
Phase II
27
35
RHT/array-system
70 MHz
Intraoperative IHP
Control
Phase II
108
Follow-up
(46)
(47)
(48)
(49)
(50)
(51)
(52)
(53)
Pediatric
nontesticular
germ cell
tumours
Cervical cancer
(recurrences)
Rectal cancer
(Dukes C
preoperative)
Metastatic Sarcoma
VP16 IFO
65
whole body
Hyperthermia
L-PAM
(dose-escalation)
ICR/2PR (19% PR)
Mitomycin C
Mitomycin C
12 NED
(best response)/
7 CR Duration:
7-64 months
5CR 2PR
(70% RR) Six
patients alive
without evidence
of tumour
(10-33 months)
2pCR/ICR 9PR
(52% RR)
3 LR
13LR
IFO/CBDCA/VP16
68% success at 1 year
P intraoperitoneal hyperthermic perfusion; WBH whole body hyperthermia; 5FU 5-flurouracil; VP16 etoposide; IFO ifosfamide;
ADR Adriamycin Doxorubin; CDDP Cisplatin; CBDCA Carboplatin; Bleo Bleomycin; L-PAM Melphan; TNF tumor necrosis factor alpha;
IFN interferon gamma; p pathohistological; RR response rate; CR complete remission; PR partial remission; MR minor response;
FHR favorable histological response >75%; LR local recurrence; NED no evidence of disease.(Published with permission from Ref. 20).
RF
input
(a)
Cooling
water
outlet
I
Io
Intensity
Cooling
channel
Piezoeletric
element
25 cm
71
I = Ioe mx
Cooling
water
inlet
Cooling
baffle
Depth, x
(b)
Latex
membrane
10 cm
Figure 6. A cross-sectional diagram of a single planar element

hyperthermia applicator. The chamber between the latex membrane and the piezoelectric element contains cooled degassed water.
During local hyperthermia treatment an ultrasonic conducing gel is
applied to the target site that is then coupled to the latex membrane.
can be directed to just that target segment. The temperatures can be adjusted in relation to variations in temperature distribution due to blood flow, variations in target
tissue morphology, and based on the patients comfort
level. These devices have the ability to contour the energy
field to match the tumor outline. These systems generally
have two frequencies: 1 MHz (used to heat 36 cm) and
3.4 MHz (used for more superficial 23 cm) (7,62). Examples of heating temperatures for different ultrasound
hyperthermia devices are shown (Table 6). These planar
array systems have been adapted to allow for thermoradiation in conjunction with an external beam radiation (7). An
extended bolus configuration with an internal reflecting
system was created to direct the ultrasound energy into
desired tissue. This configuration allows the ultrasound
transducer to be outside the radiation beam thus preventing potential interference of the two (7,70).
Another approach to achieving greater spatial control is
to use a larger variety of small transducers in a nonplanar
geometric configuration. This approach has been used in
Temperature
rise
Retaining
O ring
No cooling
Skin cooling
Depth, x
Figure 8. (a) Ultrasound intensity is greatest at the surface.
Intensity will deteriorate exponentially due to attenuation as the
depth from the surface increases. Published with the permission of
(3). (b) Since temperature and intensity are directly proportional,
temperature will decrease exponentially as depth increases. Cooling
the skin will cause the hot spot to shift to the poorly perfused fatty
tissue. (Published with permission from Ref. 3).
treating intact breast with cancer (7,71). The patient lies

prone while the breast is immersed within the water filled
cylindrical applicator (Fig. 12). The cylindrical applicator is
composed of eight rings (each ring is 25 cm in diameter by
1.6 cm in height), with up to 48 transducers (1.5 1.5 cm
plane transducers), which are interspersed around the
Skin
Necrotic/
ischemic core
Plane
wave
energy
Tumor
Well-perfused
margin
Tissue
Tumor
Figure 7. The pattern of ultrasound delivery via plane wave

radiation targeting a tumor that is located deep within the
tissue. (Published with permission from Ref. 3).
Figure 9. Temperature distribution in a subcutaneous tumor

by plane wave transducer. The temperature at the necrotic zone is
higher than in the surrounding tissues. (Published with permission
from Ref. 3).
Intensity
72
Figure 10. Pattern of radiation with a focused ultrasound beam.

The contoured beam localizes the tumor within the focal area while
sparing the surrounding tissue (Published with permission from
Ref. 3).
Depth
Temperature
rise
ring. The frequency ranges from 1.8 to 2.3 and 4.3

4.8 MHz. The driving frequency and the power can be
individually selected within each ring, which allows for
better spatial and temperature control. This technique has
not yet reached widespread clinical use (7).
The Scanning Ultrasound Reflector Linear Array System (SURLAS), which may soon be implemented in clinical
practice allows for 3D power distribution while applying
simultaneous external hyperthermia in conjunction with
radiation to superficial areas (7,13,7277). (Fig. 13). The
SURLAS applicator consists of two parallel opposed ultrasound linear arrays that aim their sound waves to a
V-shaped ultrasound reflector that further organizes and
spreads the energy over the scanned target site (7,13). The
two arrays operate at different frequencies (1.9 and 4.9).
This allows for control of penetration depth through the
exploitation of intensity modulation of the two beams (13).
The applicator housing this transducer and the temperature regulated water bolus are placed on the patient. This
system allows both the radiation and the ultrasonic waves
to enter the patients body concurrently. During the scanning interval, power levels and frequencies in each transducer can be individually regulated, thus allowing for good
control over depth penetration and lateral heating (7).
This system can treat superficial tumors that are
15 15 cm in area and with distal margins up to 3 cm
deep (13). However, scan times must be limited to <20 s to
avoid transient temperature variations >1 8C (7,73).
Large superficial tumors ranging from 3 to 4 cm deep
20 20 cm in surface area have been successfully treated
with mechanically scanned planar transducers with 2D
Depth
Figure 11. (a) With the use of a focused field, higher intensities
can be achieved in target tissue at a greater depth. (Published with
permission from Ref. (3)). (b) Since temperature and intensity are
directly proportional greater temperatures can also be attained in
target tissue at greater depths. (Published with permission from
Ref. 3).
motion (7,63) (Fig. 14). This approach can be used in

treating tumors in the chest region, which often have a
heterogenous thickness and are situated close to bone.
Once an ultrasound is launched into tissue, it cannot leave
the body; consequently, it will just bounce around until it
is completely absorbed. If the ultrasound is absorbed by
bone or nerves, neuropathies and bone necrosis can occur.
Mechanically scanned planar transducer frequencies can
range from 1 to 6 MHz. Accurate spatial control has been
achieved by controlling the operating frequency and
Table 6. Examples of Clinical Temperature and Response Rates of Certain Hyperthermia Systemsa
Device
Scanned ultrasound
Multielement ultrasound
Transrectal ultrasound
a
Reference
(63)
(64)
(65)
(66)
(67)
(68)
(69)
Number
of Patients
5
149
72
17
15
147
14
Maximum
Temperature, 8C
Minimum
Temperature, 8C
45.9
41.1
44.4
43.1
44
42.7
43.2
40.0
39.9
40.4
38.5
40.5
Average
Temperature, 8C
Complete
Response
Rate, %
34
22
24
42.3
40.4
42.2
Partial
Response
Rate, %
36
40
70
Intra-operation
cone/cylinder
Radiation
73
Patient
Tumor
Reflector and
compensator
Circumferential
ultrasound array
Degassed water
Plastic membrane
Multi- transducer application
X-Y-Z translation
& Frequency modulation
3.9 MHz
3.9 MHz
1.0 MHz
1.0 MHz
Imaging transducer
Degassed temperature
controlled water
Multiple planar transducers
Figure 14. A schematic diagram of mechanically scanned

ultrasound system for superficial hyperthermia. (Published with
Tumor
Figure 12. A schematic diagram of an intraoperative multielement

transducer system with a circumferential transducer array and
reflector configuration. (Published with permission from Ref. 7).
applied power levels, as a function of location, to account

for variations of tumor thickness. Separate transducers,
which are driven at different frequencies or by time multiplexing the driving frequency of a given transducer
between its fundamental and odd harmonic frequencies,
are able to create a situation that allows control over
penetration depth (7). The penetration depth, as well as
the temperature distribution resulting as a function of
depth, can be controlled online during the scanning by
regulating the frequency amplitude. In the clinical setting,
all these biophysical properties must be coupled with the
patients ability to tolerate the treatment to create a functional algorithm (7,63).
Scanned focus ultrasound systems (SFUs) provide the
most flexibility for clinical applications (7,6467,78,79).
These systems provide the greatest possibility of overcoming the challenges of tissue heating. The SFUs systems
generally use four to six 1 MHz spherically focused transducers each overlapped so that a common focal zone of
3 mm o.d. to treat deep tissue. This focal zone is mechanically scanned in circular or octagonal patterns within the
tumor at rates of 20100 mms1. In order to guarantee
that there is a consistency in temperature, scan cycles must
be shorter than 10 s. During scanned focused ultrasound
hyperthermia treatments, temperature distributions can
be controlled by utilizing the measured temperatures to
vary the power output as a function of the location. The
resolution is determined by a variety of thermometry
points, scanning, and computer control speed (7). The

regulation of temperature can be controlled by the clinician
or the computer (7,80).
External applicator systems for hyperthermia have now
been developed that use electrically phased focused transducer arrays. The advantages of using an electrically
phased focused transducer array is that it allows for better
synthesis of complex beam patterns and the ability to
electronically focus and steer. The 3D complex beam-forming techniques result in higher scanning speeds, smaller
applicators, and better reliability due to more static parts
(7). Examples of electrically phased focused transducer
arrays include concentric ring arrays (7,81), sector-vortex
phased arrays (7,82), spherical and cylindrical arrays
(7,83,84), and tapered phased arrays (7,85).
Intracavitary Techniques
Conventional ultrasonic hyperthermia can be used for
intracavitary applications. This modality can be used to
treat tumors that are situated deep within the body or
with those that situated close to a body cavity. Clinically,
prostrate cancer and benign prostate hyperplasia are the
best suited for this treatment (7). The transrectal applicator consists of one-half cylindrical transducer segments
1020 mm o.d. 10 mm long. It is sectored for better
angular control with frequency range of 1.01.6 MHz.
The transducers are housed in a plastic head; also, a
temperature regulated degassed water within an extendable bolus is attached (7,8688) (Fig. 15). The heating
energy is emitted radially from the length of each transducer segment, and the power is applied along the length of
the applicator. This technique is able to heat tissues that
are 34 cm deep from the cavity wall. The temperature
controlled water bolus maintains a safe temperature for
Linear accelerator
Radiation beam
Scanning reflector
High-frequency
transducer array
Low-frequency
transducer array
1/2 Cylindrical transducer segments

10-20 mm OD x 10 mm long
sectored for angular control
Expandable bolus
frequency=1.01.6 MHz
Temperature regulated
Degassed water
Tumor
Figure 13. A schematic diagram of a multielement low profile

scanning reflector system. (Published with permission from Ref. 7).
Figure 15. A nonfocused multielement applicator with

longitudinal and angular power deposition abilities. This device
is used in the treatment of the prostate cancer or BPH. (Published
with permission from Ref. 7).
74
Catheter-cooled applicator
RF feed lines
Piezooeramic tubular transducers

(1.01.5 mm OD x1020 mm long)
Temperature regulated
degassed water
RF power connection
& air- cooling ports
Water flow
13 G Implant catheter
Output port Input port
Return flow channel
Direct-coupled applicator
Biocompatible plastic
Tubular transducer
Catheter
(1.82.5 mm OD x 10 mm)
Radiation source
Thermocouple
Thermocouple monitoring
Acoustic waveguide applicator

Radiating tip
Plastic cladding
(0.61.7 mm OD x 1015 mm long)
1624 G Needle
Planar
transducer
Tapered velocity
transformer
Air gap
Figure 16. Interstitial hyperthermia catheters. (Published with

the rectal mucosa. Improved versions of this applicator

have added four sectors on each tubular section for
16 channels total. These devices are being fabricated to
be compatible with MRI guide protocols (7,89).
Interstitial Techniques
Interstitial techniques are used for treating deep-seated
and/or large tumors that are not amenable for surgical
resection. Heating sources are implanted into the tumor,
thus focusing the energy directly to the site. Interstitial
radiation is a standard practice in the treatment of tumors,
therefore incorporating adjuvant heat is a logical progression to maximizing treatment. There are three basic designs
of interstitial ultrasonic applicators: catheter cooled and
direct coupled that consists of tubular piezoceramic transducers, and acoustic waveguide antennas (7) (Figs.16ac).
Multielement ultrasound applicators with catheter
cooling use piezoceramic tubular transducers (1.01.5 mm
o.d. 1020 mm long, with frequency ranging from 7 to
10 MHz) have circulating coolant channels incorporated
within the support structures to allow the applicator to
be sealed in place within closed end implant catheters
(1314 gauge) (7) (Fig. 16a). These catheters are able to
improve control of radial penetration of heat. In addition, it
has the ability to control longitudinal power deposition
along the length of the applicator (7,9094). The power to
each tubular transducer can be adjusted to control tissue
temperature along the length of the catheter. The length
and the number of transducers can be selected depending on
the desired temperature and longitudinal resolution. This
feature is very valuable in that it allows adjustability to
tumor geometry variations, blood perfusion variations, and

the variation within the tumor tissue. Another advantage
of this device is that, unlike microwaves and RF hyperthermia, the power deposition pattern is not limited by the
length of insertion or whether other catheters are within
the implant. These catheters are more challenging than
others for the operator to use skillfully because it is complicated to control both the electronics and the water cooling.
Also, small transducers are less reliable. However, it is this
complexity that allows for great plasticity in therapeutic
temperature distributions (7).
Direct coupled applicators are used to deliver thermobrachy therapy via remote afterloading radiation sources
(Fig. 16b). Larger applicator size limits these catheters to
few clinical treatments. The implant catheter consists of
the transducer and an acoustically compatible housing,
which is biologically and electrically insulated. The
implant catheter usually ranges from 2.2 to 2.5 mm in
diameter. The inner lumen is formed from a catheter that
is compatible with standard brachytherapy and commercial after loaders. The transducers have sensors that are
able to monitor tissue temperature. In order to conserve
size, a water cooling mechanism was not included as part of
the catheter. This device is less efficient because transducer self-heating increases the wall temperature and thus
reduces radial heating. Therefore, the thermal penetration
is sensitive to acoustic frequency (7,95,96). Some studies
have shown that integrating an air cooling system to this
catheter will allow for better heating penetration (7,95).
The acoustic wave-guide antenna has a minimally invasive 1624 gauge stainless steel needle that is coupled by a
conical tapered velocity transformer to a piezoceramic disk
transducer (1.3 cm o.d. operating at 1 MHz) (7,99)
(Fig. 16c). The length of the radiating tip can be changed
by adjusting the length of the plastic sleeve by 11.5 cm.
The needle diameter size minutely fluctuates due to
Raleigh surface waves propagating from the wave-guide
generating flexural vibrations of the needle portion. Acoustic patterns that have been measured demonstrate peaks
and nodes in adjacent tissue along the radiating aperture.
The temperature of the tissue that is radiated matches the
temperature of the radiating antennae. The disadvantages
of this system are that the power output is potentially
limited for larger or more perfused tumors, and it is
difficult to control the longitudinal power deposition (7).
Focused
ultrasound
transducer
Skin
Tissue
target
Focal
volume
Focused
ultrasound
transducer
Tissue
target
Skin
Figure 17. Schematic of HIFU. (a) Illustrates a formation of a

single lesion. (b) Illustrates a confluent array of lesions required
for a therapeutic effect. (Published with permission from Ref. 98).
Figure 18. Image of coagulation and liquefied necrosis created

with HIFU in an ex vivo porcine kidney. (Published with permission
from Ref. 108).
A Brief History of HIFU

Using HIFU as an extracorporeal technique of creating
coagulative necrosis was first conceptualized in 1942 by
Drs. Lynn and Putnam (12,98,99) (Fig. 17a). In 1954, Dr.
William Fry was the first to use HIFU to destroy central
nervous tissue in the brains of cats and monkeys
(12,98,100,101). Later, Frank Fry treated patients with
Parkinsons disease and neuromata (12,98,102). Throughout the 1950s and 1960s, HIFU research continued,
although it was often plagued with limited success due
to lack of technology (103106). In 1956, Dr. Burov suggested that HIFU can be implemented in the treatment of
cancer (12,98,107). Since then, the popularity of HIFU has
gradually increased with the advent of better devices and
with the success of its use in vitro and in vivo experimental trials. In current literature, HIFU is categorized
as a high temperature hyperthermia because higher temperatures than those used in conventional hyperthermia
are required to achieve therapeutic goals.
BASIC PRINCIPLES OF HIFU
The concept of HIFU is similar to that of using a magnifying glass to focus the suns beams to set fire to some dry
leaves. Only the leaves that are in focus will be set on fire,
the surrounding ones will be spared (12,98). Likewise, if an
ultrasonic beam with sufficient energy is tightly focused, it
can be used to elevate temperatures within a target tissue
resulting in cell death and coagulative necrosis while
sparing the skin and surrounding tissues (98,108)
(Fig. 18). Histologically, there is a sharp demarcation
between the necrotic tissue that was radiated with HIFU
and the healthy surrounding tissue. In the liver, 2 h after
exposure, the cells look normal, however, approximately a
10 cell wide rim of glycogen poor cells can be found. After
48 h, the entire area that was radiated will be dead (109).
During HIFU procedures, tissue temperature >56 8C
are used because at that temperature irreversible cell
75
death through coagulative necrosis occurs. The main

mechanism used is coagulative necrosis via thermal
adsorption (110). The other mechanism is cavitation
induced damage that is caused by both thermal and
mechanical properties of the ultrasound wave (110,111).
However, recent studies have been investigating the use of
cavitation to enhance the level of ablation and to reduce
exposure times. It has been proposed that a focused ultrasound protocol that induces gas bubbles at the focus will
enhance the ultrasound absorption and ultimately create
larger lesions (110,112). Individual HIFU exposure times
can be as little as 13 s, while larger volumes may require
up to 3060 s. Individual lesions can be linearly complied to
create a clinically relevant lesion (Fig. 17 b). Since individual exposure time is quick, issues (e.g., the cooling
effects of blood perfusion) can be considered negligible
(7,98,113,114). Therefore, energy transfer and temperature elevation in tissue is considered proportional to acoustic field energy (100). The lesions are cigar-shaped or
ellipsoid with the long axis parallel to the ultrasonic beam
(12,98). In order to ablate tissue transducer frequency
must be between 0.5 and 10 MHz. The higher the frequency, the narrower and shallower the lesion will be.
The wavelength ranges from 3 to 0.25 mm. The size of
the focal point is determined by the wavelength. Thus, the
transverse diameter of the focus is limited to one wavelength and the axial diameter is eight times that wavelength. As a result of this, all generators create a focal size
that is 10 1 mm. The shape of the lesion is determined by
the acoustic properties of the tissue, ultrasound intensity
in conjunction with exposure time, and transducer geometry (12). Lesion size is determined by power density at the
focus, pulse duration, and the number of pulses. In order to
create a well-demarcated lesion the intensity must be
>100 Wcm2, thus being able to reach temperatures that
are >65 8C in <5 s (11). Focal peak intensities generally
range between 300 and 2000 Wcm2 (7). The ultrasonic
waves used in HIFU are generated by piezoelectric elements. In order to achieve high intensity focus ultrasound
that is able to ablate tissues three techniques have been
found to focus the ultrasound beam: (1). spherical arrangement of piezoelements (Fig. 19), (2) combination of a plane
transducer with an acoustic lens (Fig. 20), (3). cylindrical
piezoelements together with a parabolic reflector (11) (Fig. 21).
CURRENT EXTRACORPOREAL DEVICES, INTRACAVITARY

DEVICES, AND IMAGING
While there are many devices that are used in experimental trials, few of those are currently used in widespread
clinical practice. The two main categories of HIFU devices
are extracorporeal and transrectal. Extracorporeal devices
have been implemented in experimental trials in many
medical fields. Extracorporeal devices use larger transducers, lower frequencies, and longer focal lengths than
intracavitary devices (97).
An important factor in clinical application of these
devices is the ability to monitor treatment accurately.
In current practice, this is accomplished either by using
real-time ultrasound (116118) or MRI (119122). When
76
Focal point 32x4mm

Treated volume
Focal distance
100 mm
Beam path
Transducer
1 MHz
Ultrasound field
Piezoelectric
cylinder
Parabolic
reflector
Figure 19. A single spherically curved focused transducer.
MR is used to guide HIFU treatments, sublesioning ultrasound exposures are used to identify the target region, local
rise in temperatures are used to confirm the position of the
ultrasound focus and then higher intensity therapeutic
exposures are used for treatment. Currently, several
groups are using ultrasound surgery systems that utilize
MRI to map temperature elevations online during HIFU
procedures (110,120122). This technique has been used to
treat breast tumors and uterine fibroids, and these treatments are in the process of being used clinically in several
countries (110,123125). The MR can effectively use temperature data to determine the parameter of thermal tissue
damage (110) and is limited in that it is costly, has lower
spin resolution, and because of its technology for producing
MR compatible ultrasound equipment required for HIFU is
lagging.
When ultrasound is used as a guide, the diagnostic
transducer is arranged confocally with the therapeutic
transducer and their relationship is fixed. The position
100 mm
B-mode
ultrasound probe
Figure 21. A cylindrical transducer with parabolic reflector.

of the therapeutic focus can be reliably identified on the

diagnostic image. The extent of treatment can be monitored by recording post-treatment gray scale changes on
the diagnostic image (98). Ultrasound as a guide is advantageous in that it is less expensive and is more readily
accessible, it has faster treatment times, compact sized
equipment, and provides a good correlation between
observed ultrasound changes and the region of necrosis
in the tissue. The disadvantage of using ultrasound as a
guide is that image quality is not optimal (98,110,126).
Furthermore, ultrasound waves are obstructed by bone
and air-filled viscera.
MEDICAL APPLICATIONS OF HIFU
Aluminum lens
Latex condom
Brass
housing
Water tubing
External view of HIFU

transducer
Figure 20. A plane transducer with an acoustic lens used for

focusing the ultrasound waves. (Published with permission from
Ref. 115).
Liver Cancer
While hepatocellular carcinoma is frequently encountered in
clinical practice, hepatic metastasis from other primary
sources is much more common. Currently, the only definite
treatment choice for hepatic metastases is surgery, however,
5 year survival rates are only 2530%. Arterial embolization
is another emerging technique. Therefore, the desire to find a
noninvasive technique is preeminent (98). The Chongqing
HAIFU device has been used for a couple of years in China to
treat a variety of tumors, however, adequate data has not yet
been collected (98,127,128) (Fig. 22). The JC-HIFU system
(HAIFU Technology Company, Chongquing, PR China) uses
an extracorporeal transducer that operates at 0.81.6 MHz,
the aperture 1215 cm, focal length 915 cm. It operates at
Isp of 515 kWcm2. A diagnostic ultrasound probe
(3.5 MHz) is aligned along the same axis as the therapeutic
transducer. Both the treatment and diagnostic transducers
are placed in a reservoir of degassed water in the center of the
77
Figure 22. A Hifu System that is used both clinically and experimentally in the treatment of liver
metastates. (Published with permission from Ref. 128).
treatment table. The degassed water provides acoustic coupling between the patient and the transducer. Horizontal
movement of the transducer is possible along three orthogonal axes of the bed because it is facilitated by the cylindrical
gantry at one end of the table. All movement is controlled by
the adjacent computer terminals (128). In a recent clinical
trial carried out in Churchill Hospital in Oxford, England in
conjunction with Chongqing University of Medical Sciences
in Chongqing China, 11 patients with liver metastases were
treated with the JC-HIFU device. While it is not possible to
have a good statistical analysis with such a small subject pool,
some general observations were made about the safety of this
device. Of the 11 patients treated, 7 out of 11 patients
complained of transient pain and 3 out of 11 complained of
superficial burns. Out of the 7 patients that experienced pain,
oral analgesia brought relief to 6. Burn sites were treated
with ice-packs and aloe gel. Two of the three burn sites were
only millimeters across. One of the burns was 2 3 cm and
had healed by the 2 week follow-up period. It appears that
from a safety standpoint the JC-HIFU is a feasible treatment
option for hepatic metastases, however, larger trials will be
needed to determine the true efficacy of the treatment (128).
Another study by Wu et al looked at 55 patients with
hepatocellular carcinoma with cirrhosis. Tumor size ranged 414 cm in size with an average size of 8.14 cm.
Patients were classified according to progression of disease:
15 patients had stage II, 16 had stage IIIA, and 24 had
stage IIIC. All patients were treated with an extracorporeal HIFU device similar to the one previously mentioned
for the treatment of liver metastases. The average number
of treatment applications was 1.69. There were no serious
side effects. Imaging following HIFU treatment evaluated
for the absence of tumor vascular supply and shrinkage of
treated lesions. Serum alpha-fetoprotein returned to normal
in 34% of patients. At 6 months, 86.1% of the patients were
still alive, at 12 months 61.5% of the patients were still alive,
and at 18 months 35.3% of the patients were still alive. The
survival rates were the highest in patients who were stage
II. Therefore, this study demonstrated that HIFU is a safe

option in the treatment of hepatocellular carcinoma (129).
Prostate Cancer
Prostate cancer is one of the common types of cancer in
males, and it is frequently the cause of cancer-related
death (130). Since physicians are able to detect prostate
cancer early, there has been an increase in the number of
patients needing treatment. Radical prostatectomy is the
treatment of choice in patients who have organ-confined
disease and a life expectancy of >10 years. Radical prostectectomy offers excellent results 5 and 10 years after the
operation, although there is still risk of morbidity associated with the operation, thus precipitating the need for
a noninvasive procedure. Currently, brachytherapy, cryosurgery, 3D conformal radiotherapy, and laparoscopic
radical prostatectomy have been implemented with good
results (130,131). However, if a cure is not achieved, these
treatments cannot be repeated and there is high risk of
morbidity associated with salvage radical prostatectomy,
thus necessitating the need for another treatment option.
In 1995, Madersbacher reported that they were able to
destroy the entire tumor within the prostate (98,132). Early
reports showed success rates of controlling local tumors at
50% at 8 months and then approaching 90% in later studies
(98,133,134). In the later years, as clinicians gained more
experience and as technology has improved, treatment of
the entire gland was performed (98,135,136).
A recent report was published that looked at 5 year
results with transrectal high intensity focused ultrasound
in the treatment of localized prostate cancer. One hundred
and forty six patients were treated with Ablatherm device
(EDAP, Lyon, France). The tablespoon-shaped intracavitary applicator contains both a 7.5 MHz retractable ultrasound scanner for diagnosis and a piezoelectric therapeutic
transducer that can be driven at frequencies of 2.25
3.0 MHz. The computer-controlled treatment head is able
78
to move three dimensionally. The applicator can be rotated

45 8 in the rectal ampulla. A cooling device that consists of
a balloon containing degassed coupling fluid surrounds the
treatment head. Energy can be released from the balloon
rectal interface thereby maintaining rectal temperatures
<15 8C. Out of 137 patients 6 reported symptomatic UTI, 2
reported chronic pelvic pain, 16 reported infravesicular
obstruction, 8 reported grade1 stress incontinence, and 1
reported rectourethral fistula. The success rate of the
Ablatherm system is between 56 and 73% (131).
Another study that was published by Uchida et al.
performed 28 HIFU treatments on 20 patients to treat
localized prostate carcinoma (T1b-2NOMO). A modified
Sonoblate 200 HIFU device (Focus Surgery, Indianapolis
Ind.) was used in this study. Sonoblate 200 uses a 4 MHz
PZT transducer for both imaging and treatment. Each
pulse delivery ablates a volume of 2 2 10 mm3 in a
single beam with 2.5 and 4.5 cm focal length probes. Probes
with focal lengths of 3.0, 3.5, 4.0 cm can be used in a splitbeam conformation to create lesion sizes of 3 3 10 mm3.
A cooling device maintains rectal temperatures at <22 8C.
In this study, there was a 100% success rate. The UTI-like
symptoms were common in the first 2 weeks post-HIFU, but
were easily remedied with alpha-blockers and painkillers.
One patient had a rectourethral fistula after a second HIFU
treatment. Of 10 patients who were still able to attain
tumescence prior to the procedure, 3 reported postoperative
impotence. It is hypothesized that the reason the Sonoblate
200 is getting superior results to the Ablatherm system is
that the treatable focal length is longer in the Sonoblate
system. This allows the Sonoblate 200 to treat prostates
<50 mL, whereas the Ablatherm can only treat prostates
that are <30 mL. However, a controlled prospective study
is needed to evaluate the potential reasons for this difference in efficacy (130).
Gynecology
The most common pelvic tumor in women of reproductive
age is fibroids. The current surgical options available to
manage fibroids are either hysterectomy or myomectomy.
Hysterectomy is often not a viable option for women who
wish to have children. Myomectomies often result in 50%
tumor recurrence in 5 years. Hormone therapy results in
temporary reduction in tumor size by 3565% (115). Therefore, there is a need for a permanent, noninvasive technique to manage fibroids. A device was developed for treating
uterine fibroids. The prototypic device aligns a commercial
abdominal diagnostic ultrasound transducer with a therapeutic ultrasound intracavitary probe (Fig. 23). This
device was constructed to accommodate the specific constraints of the female pelvic anatomy. The transducer
contains a 3.5 MHz PZT-8 crystal, 25.4 mm in diameter
bonded by an aluminum lens to focus the ultrasound beam.
A water-filled latex condom is used for acoustic coupling of
the transducer and also has the potential for transducer
and tissue cooling. Ergonomics testing in humans demonstrated clear visualization of the HIFU transducer in relation to the uterus, thereby demonstrating a potential for
HIFU to treat fibroids from the cervix to the fundus
through the width of the uterus. However, this device
Ultrasound
image probe
Image
Water plane
filled
condom
Ut
er
Bla
Vagina
dd
us
Device
er
Transvaginal
HIFU transducer
Rectum
Figure 23. Conceptual diagram of the transvaginal device HIFU

in use. The dotted line represents the image plane. Three factors
determine what area can be treated by the HIFU transducer: focal
length, the range of water stand-off, and the range of mobility once
inside the vagina. (Published with permission from Ref. 115).
needs to be tested in treating the uterus in large animal

models prior to beginning human trials (115). Extracorporeal devices have been used in a small clinical phase I trials,
but the results are still pending (98,137).
Breast Cancer
Every year >1 million new cases of breast cancer are
diagnosed. Breast cancer is the most common malignancy
in women (138,139). In the past, the only options available
to women were radical and modified radical mastectomy
that included dissection of the axillary lymph nodes. More
recently, breast conservation surgery in conjunction with
radiotherapy, chemotherapy, and hormone therapy has
gained popularity in early stages of breast cancer. However, the change in approach toward a less radical surgery,
while being better for a womans body image, has not really
increased long-term survival rates in breast cancer
patients (138,140). Other options, such as cryoablation
and laser frequency have been studied as minimally invasive approaches, however, these techniques are limited in
that they require percutaneous access and that they are
only able to treat small masses. In a recent study by Wu et
al., (65) women with breast cancer (T1-2, N0-2, M0) verified
with biopsy were studied. Patients were divided into a
control group that had a modified radical mastectomy,
and a group that had extracorporeal HIFU followed by a
radical mastectomy. The HIFU system used is the same
one described earlier in the treatment of liver malignancies, the JC-HIFU therapeutic system. The therapeutic
U.S. beam was produced by a 12 cm in diameter PZT-4
ceramic transducer with a focal length of 90 mm that was
driven at a frequency of 1.6 MHz. The ellipsoid focal region
dimensions were 3.3 mm along the beam axis, and 1.1 mm
along the transverse axis. A real-time imaging U.S. device,

the AU3 (Esaote, Genoa, Italy) was used at frequencies of
3.55.0 MHz. The diagnostic transducer is placed in the
center of the therapeutic transducer. Real-time imaging
can accomplish three separate tasks. It can locate the
tumor that needs to be treated, it can guide the deposition
of U.S. energy into the tumor, and it can provide real-time
assessment of the coagulation necrosis during therapy.
The results demonstrated that there were no severe side
effects. Those that were reported included mild local pain,
warmth, and a sensation of heaviness in the affected
breast. However, only 4 of the 23 HIFU patients had
significant pain to require a 35 day course of oral analgesics. Only one patient had a minor skin burn. Pathologic
examination of the breast tissue revealed complete coagulative necrosis, and the tumor vasculature was damaged.
The immunohistological staining revealed that no expression PCNA, MMP-9, and CD44v6 was found, indicating
that the tumor cells had lost their ability to proliferate,
invade, and metastasize. Therefore, this study demonstrated the safety and efficacy of HIFU in the treatment
of breast cancer (138).
Neurology
Recently, there have been some published studies that
propose using large array ultrasound transducers to overcome distortions caused by the skull (110). The goal has
been to be able to create an array that can focus to destroy
target tissue while preserving surrounding tissue. A 320
element array has been used to focus ultrasound through10
human skulls. This approach is completely noninvasive.
This technique is modeled after a layered wave vectorfrequency domain-model and uses a hemisphere-shaped
transducer to propagate ultrasound through the skull
using CT scans as a guide (110,141,142). The ability to
focus energy has implications that are not limited to just
tumor treatment. It has been shown that focused ultrasound can selectively and consistently open the blood brain
barrier (BBB) (110).
Another neurological area that may benefit from HIFU
is in the treatment of nerve spasticity and pain. Spasticity, which is signified by uncontrollable muscle contractions, is difficult to treat. In a recent study, HIFU was
used to treat and suppress the sciatic nerve complex of
rabbits in vivo. An image-guided HIFU device including a
3.2 MHz spherically curved therapeutic transducer and
an ultrasound diagnostic device were used. A focal intensity of 14801850 W/CM2 was used to create a complete
conduction block in the 22 nerve complexes. Treatment
times averaged 36 s. Gross histological examination
revealed blanched nerve complex with lesion dimensions
of 2.8 cm3. Further histological examination revealed the
probable cause of nerve block as axonal demyelination
and necrosis of Schwann cells. This study illustrates the
potential that HIFU may have in the treatment of nerve
spasticity (143).
Cardiovascular System
The role of ultrasound in cardiology has been instrumental
to the increasing knowledge of the cardiovascular system.
79
Diagnostic ultrasound technology has led to a greater

understanding of the anatomy and physiology of the
human heart and vascular systems. Over the years, in
addition to diagnostic use, the role of ultrasound has been
expanded to the therapeutic realm. Both conventional
ultrasound and HIFU modalities have been used with
varied success in many cardiovascular therapeutic applications. These applications range from harvesting the
internal mammary artery for coronary artery bypass surgery to ablation of cardiac arrhythmias. An ultrasonically
activated (vibrating up to 55,000 Hz) harmonic scalpel
(Ethicon Endosurgery, Cincinnati, OH) produces low heat
(<100 8C) thereby effectively coagulating and dividing the
tissue and has a wide range of applications in cardiothoracic surgery (144). By using a 1 MHz phased array transducer with an acoustic intensity of 1630 Wcm2 or 22547
Wcm2 one can successfully create precise defects ranging
from 3 to 4 mm in diameter ex vivo in porcine valve leaflet,
canine pericardium, human newborn atrial septum, and
right atrial appendage (145). Cardiac arrhythmia is one
area where significant work has been done using ultrasonic
hyperthermia for therapeutic purposes. Strickberger et al.
demonstrated an extracorporeal HIFU ablation of the
atrioventricular junction of beating canine heart after
thoracotomy (146). Their experimental system consisted
of a polyvinyl membrane covering the heart and lungs. The
thoracic cavity was filled with degassed water serving as a
coupling medium. A 7.0 MHz diagnostic 2D ultrasound
(Diasonics VST Master Series, Diasonics/Vingmed Ultrasound Inc) attached to a spherically focused single piezoelectric element therapeutic ultrasound transducer
(1.4 MHz frequency; 1.1 8.3 mm focal length and
63.5 cm focal zone) with the maximum intensity of
2.8 kWcm2 was applied during the diastole for 30 s to
achieve complete AV nodal junctional block (Fig. 24ac).
Experience with HIFU application is very preliminary and
has not been tried for AV nodal ablation in humans yet.
Ultrasound had also been used clinically in the treatment of atrial fibrillation (AF), which is the most common
arrhythmia affecting 0.52.5% of the population globally.
Over the last decade, ablation procedures by isolating the
pulmonary veins and eliminating electrical triggers from
the atria has become a popular and effective mode of
therapy for AF. Traditionally, RF energy has been used
as an energy source for ablation. Radio frequency catheter
ablation of AF requires good tissue contact, multiple
lesions, significant experience and manual skills with long
procedure time. Complications related to RF application in
AF ablation include pulmonary vein stenosis, atrioesophageal fistula, left atrial rupture due catheter perforation or
inappropriate amount of power. The limitations of the
existing RF technology could be overcome with the use
of HIFU balloon systems (147).
Our group performed the initial work on pulmonary
vein isolation in humans using a through-the-balloon circumferential ultrasound (conventional-unfocused) ablation system for treatment of recurrent atrial fibrillation
(148). Fifteen patients with drug refractory atrial fibrillation underwent a PVI using a novel transballoon ultrasound ablation catheter (Atronix, Inc) (Fig. 25ac). The
ablation system was composed of a 0.035 in. diameter
80
Figure 24. (a) Schematic of the experimental HIFU apparatus. The therapeutic ultrasound
transducer is mounted 63.5 mm from the target (X). A polyvinyl chloride membrane covers the
heart and the lungs. Degassed water flows in and out of the thoracic cavity at a rate of 600 mLmin1.
Combined diagnostic/ablation transducers are placed into degassed water. (Published with
permission from Ref. 146). (b) ECG and Echocardiogram of a canine heart. Prior to ablation of the
AV node. (c) After ablation of the AV node using HIFU, the ECG shows complete AV block and the echo
image depicts an increased density of the ablated tissue. (Published with permission from Ref. 146).
81
nels were successfully created in canine myocardium (151).

This shows the potential for future application for HIFU in
TMR in a noninvasive fashion.
Other Applications of HIFU
Figure 25. Anatomical pulmonary vein variations and technical

limitations. (Published with permission from Refs. 148 and 149).
luminal catheter with a distal balloon (2.2 cm at maximum

diameter) housing a centrally located ultrasound transducer (8 MHz). The ultrasound ablation system was
advanced over a guide wire (0.035 in., 0.088 cm) to the
intended pulmonary vein. The ablation performance and
tissue temperature are monitored by thermocouples on the
balloon and the therapeutic transducer. The ablation
time was 2 min with an additional minute to deflate the
balloon. The energy was delivered perpendicularly to the
surface of the balloon and ablation at the funnel portion of
the pulmonary vein (antrum) could not be achieved with
the original design. Additionally, other anatomical characteristics of the target sites like ostial diameter larger
than the balloon size, inability to reach the right inferior or
other pulmonary vein ostia, ostial instability, early branching of the vein, and eccentric position of the ultrasound
transducer in the vein made it difficult to deliver energy
effectively (Fig. 26). These technical limitations have been
addressed in some of the newer balloon systems where the
energy delivery could be accomplished in a divergent
angle enabling ablation around the antrum with the tip
of the balloon sitting at the pulmonary vein ostium. Early
animal studies on HIFU mediated AF ablation have
shown promising results with an experimental device that
focuses ultrasonic energy via a parabolic balloon, using
gas or fluid as a reflector (ProRhythm INC.) (Fig. 25ce).
(149,150).
Since 2003 60 patients were treated using this system.
With improved catheter design the success rate of AF
ablation has increased from 50 up to 80% without evidence of pulmonary vein stenosis. These preliminary
human study results need to be confirmed in larger series.
Since there are no large clinical trials on AF ablation with
this technology, it is still somewhat premature to predict if
HIFU is the complete answer. This device is expected to be
released in Europe in 2005 (149,150).
Attempts have been made to harness HIFU for transmyocardial revascularization (TMR) to improve blood supply to damaged myocardium caused by advanced heart
disease. Using a 10 cm diameter transducer operating a
frequency of 2.52 MHz, intensity of 2300 Wcm2 and pulse
repetition period of 40 ms at 50% duty cycle, small chan-
Several branches of medicine have already begun to benefit

from the use of HIFU with the prospect of many more
applications in the future. Thus far the greatest influence
of HIFU has been in oncology, with other fields now exploring and experimenting with HIFU to determine its potential utility. The HIFU has been proposed as a tool for
synovectomy in the treatment of rheumatoid arthritis
(RA) (98,152) and has been used to control opiate refractory
pain in pancreatic cancer patients (98,127) and internal
bleeding in organs and vessels (98,153). A hand-held HIFU
device has been successfully used to perform vasectomies
in dogs as a 12 min procedure (98,154).
Future Perspectives in Conventional Hyperthermia
and HIFU Use
Heat as therapeutic entity has had a rich history punctuated with many successes and failures. The evolution and
integration of therapeutic hyperthermia in the clinical
setting have been the product of clinical trials, development of new devices, and education of the medical personnel. Conventional hyperthermia has been used for a long
time with many energy sources such as electromagnetic,
ultrasound, and microwaves. Similarly, it has been >50
years since HIFU was first conceptualized and actualized.
Many subspecialties of medicine have benefitted from the
use of hyperthermia. Some of the limitations that were
miring conventional hyperthermia and HIFU have only
recently begun to be overcome and now these therapies can
reach a wider patient population. Both of these techniques
share similar obstacles due to the limitations that are
inherent to ultrasound. Indeed, ultrasound hyperthermia
procedures are limited in that ultrasound cannot propagate through air-filled cavities (e.g., lung or bowel). Consequently, lung tumors other than those that are at the
periphery are not likely to be amenable to treatment with
HIFU or conventional hyperthermia. Also, tumors that are
in close proximity to the bowel or within the bowel wall
pose an increased chance of visceral perforation with HIFU
use. In addition, other side effects such as pain, soft tissue
and bone damage, and skin burns have been reported.
Often these side effects can be minimized by varying scan
paths, altering frequency, power deposition, or the applicator position (7). The success of both hyperthermia techniques is determined by whether ultrasound energy can be
properly directed to the site of interest, or if therapeutic
temperatures are achieved, and if other factors can be
compensated for, such as hemodynamic changes. Other
challenges facing ultrasound hyperthermia systems
include the ability to gain control over spatial distribution
of heat, tissue temperature monitoring, and improved
diagnostic visualizations in order to better treat the tumor
site. The HIFU treatment times also need to be shortened.
Despite this, treatment times of 1 h for a 2 cm superficial
tumor using HIFU is preferable to surgical resection; conversely, at present the same tumor can be treated much
82
Figure 26. (a) A schematic of the first ultrasound balloon system illustrating the radial delivery of
ultrasound energy that is transferred perpendicularly to the tissue. (b) 8F transballoon ultrasound
ablation catheter with a central lumen that can accommodate a 0.035 in. guidewire. The distal end of the
catheter lodges a cylindrical transducer axially with a saline-filled balloon inflated over it. (c) The
balloon is advanced over a guidewire at the ostium of the left upper pulmonary vein. An occlusive
pulmonary venogram is done to confirm that the transducer is located at the proximal portion of the
vein. Published with permission from Refs. 148 and 149. (d) A schematic of the HIFU device illustrating
a ring pattern of ultrasound transmission. The parabolic balloon focuses the ultrasound beam forward.
The HIFU catheter is able to create lesions in the antrum away from the lumen of the pulmonary veins
thus reducing the likelihood of pulmonary vein stenosis. (e) The HIFU catheter at the ostium of the
pulmonary vein.
more quickly with radiofrequency ablation. However, in

large tumors longer treatment times are justified because
there are very low mortality and morbidity rates associated
with the use of HIFU (98).
The benefits of both conventional hyperthermia and
HIFU are limitless, but both need more testing in larger
patient populations to fully delineate their clinical role.
At present, the scarcity of available equipment, the
required technical expertise and the lack of thoroughly
trained providers, the increased complexity and difficulty
involved in using conventional hyperthermia and HIFU
when compared to other systems (e.g., electromagnetic)
have limited the widespread use of this treatment modality
(7). Issues, such as accurate delivery of focused ultrasound
to the target tissue, the ability to monitor and control
temperatures, and reasonable treatment times, have all
been improved upon and will continue to be refined with
further testing and research. Imaging modalities like
MRI and high resolution CT scan will enhance the precision of HIFU in various system applications. In the future,
HIFU could emerge as a good alternative to traditional
surgery, and conventional hyperthermia may become a
mainstay as an adjunct to chemotherapy and radiation
to ultimately improve the arsenal of methodologies available to better treat patients.
BIBLIOGRAPHY
1. Seegenschmiedt MH, Vernon CC. A historical perspective
on hyperthermia in oncology. In: Seegenschmiedt MH,
Fessenden P, Vernon CC, editors. Thermoradiotherapy and
Thermochemotherapy Vol 1. Berlin: Springer Verlag; 1995.
pp 344.
2. Modern cancer treatment by oncothermia and hyperthermia.
2003. Available at http://www.hot-oncotherm.com/historical.
htm. Accessed 2005, Feb 4.
3. Lele P. Hyperthermia Ultrasonics. Encyclopedia of Medical
Devices. New York: John Wiley, Sons; 1998.
4. History of Hyperthermia. Available at http://www.starfarm.it/
hyperthermia/history.htm. Accessed 2005, April 4.
5. van der Zee J. Heating the patient: A promising approach?
Ann Oncol 2002;13:11731184.
6. Fesseden P, Hand JW. Hyperthermia therapy physics. In:
Smith AR, editor. Medical radiology: Radiation therapy physics. Berlin: Springer Verlag; 1995. pp 315363.
7. Diederich CJ, Hynynen K. Ultrasound technology for
hyperthermia. Ultrasound Med Biol 1999;25:871887.
8. Bushberg J, Seibert J, Leidholdt E, Boone J. The Essential
Physics of Medical Imaging. Philadelphia Lippincott,
Williams, and Wilkins; 2002. pp 469553.
9. Chapelon JY, et al. New Piezoelectric transducers for therapeutic ultrasound. Ultrasound Med. Biol. 2000;26:153159.
10. Moros E, et al. Thermal contribution of compact bone to
intervening tissue-like media exposed to planar ultrasound.
Phys Med Biol 2004; 869886.
11. Madersbacher S, Marberger M. High-energy shockwaves and
extracorporeal high-intensity focused ultrasound. J Endourol
2003;17:667672.
12. ter Haar G. High intensity focused ultrasound for the treatment of tumors. Echocardiography 2001;18:317321.
13. Novak P, et al. SURLAS: A new clinical grade ultrasound
system for sequential or concomitant thermoradiotherapy of
superficial tumors: Applicator description.
83
14. Reinhold H, Endrich B. Tumour microcirculation as a target

for hyperthermia. Int J Hypertherma 1986;2:111137.
15. Vaupel PW, Kelleher DK. Metabolic status and reaction to
heat of normal and tumor tissue. In: Seegenschmiedt MH,
Fessenden P, Vernon CC. editors. Thermoradiation and Thermochemotherapy, Vol. 1. Berlin: Springer Verlag; 1995. pp
157176.
16. Raaphorst GP. Fundamental aspects of hyperthermic biology.
In: Field SB, Hand JW, editors. An Introduction to the Practical Aspects of Clinical Hyperthermia. London: Taylor and
Francis; 1990. pp 1054.
17. Fajardo LF, Pathological effects of Hyperthermia in normal
tissue. Cancer Res 1984;44:4826s4835s.
18. Sminia P, et al. Effects of hyperthermia on the central nervous
system: A review. Int J Hyperthermia 1994;10:1130.
19. Wondergem J, et al. Effects of local hyperthermia on the motor
function of the rat sciatic nerve. Int J Rad Bio 1988;53:429439.
20. Falk MH, Issels RD. Hyperthermia in oncology. Int J
Hyperthermia 2001;17:118.
21. Overgaard J, et al. Randomised trial of hyperthermia as
adjuvant to radiotherapy for recurrent or metastatic malignant melanoma. The Lancet 1995;345:540543.
22. Sneed PK, et al. Survival benefit of hyperthermia in a prospective randomized trial of brachytherapy boost (hyperthermia for glioblastoma multiforme). Int J Rad Oncol Biol Phys
1998;40:287295.
23. Vernon CC, et al. Radiotherapy with or without hyperthermia
in the treatment of superficial localized breast cancer: Results
from five randomized controlled trials. Int J Rad Oncol Biol
Phys 1996;35:713744.
24. Valdagni R, Amichetti M, Pani G. Radical Radiation alone
versus radical radiation plus microwave hyperthermia for N3
(TNM-UICC) neck nodes: A prospective randomized clinical
trial. Int J Rad Oncol Biol Phys 1988;15:1324.
25. Datta N, et al. Head and Neck cancers: Results of thermoradiotherapy versus radiotherapy. Int J Hyperthermia
1990;6:479486.
26. Perez et al. Randomised phase III study comparing irradiation
and hyperthermia with irradiation alone in superficial measurable tumors. Am J Clin Oncol 1991;14:133141.
27. Jones E, et al. Randomized Trial of Hyperthermia and radiation
for superficial tumors. J Clin Oncol 2005;23(13):30793085.
28. Emami B, et al. Phase III study of interstitial thermoradiotherapy compared with interstitial radiotherapy alone in the treatment of recurrent or persistent human tumors: A prospectively
controlled randomized study by the radiation therapy oncology
group. Int J Rad Oncol Biol Phys 1996;34:10971104.
29. Harima Y, et al. A randomized clinical trial of radiation
therapy versus thermoradiotherapy in stage IIIB cervical
carcinoma. Int J Hyperthermia 2001;17(2):97105.
30. Vasanthan A, et al. Regional hyperthermia combined with radiotherapy for uterine cervical cancers: A multi-institutional prospective randomized trial of international atomic energy agency.
Int J Rad Oncol Biol Phys 2005;61(1):145153.
31. Van der Zee J, et al. Comparison of radiotherapy alone with
radiotherapy plus hyperthermia in locally advanced pelvic
tumors: a prospective, randomized, multicentre trial. Dutch
Deep Heat Hyperthermic Group. Lancet 2000; 11191125.
32. Perez CA, Gillespie B, et al. Quality assurance problems in clinical
hyperthermia and their impact on therapeutic outcome: a report
by the radiation oncology group. International Journal of Radiation Oncology, Biology, and Physics, 16:551558.
33. Voldagni R, Amichetti M. 1994 Report of long-term follow-up
in a randomized trial comparing radiation therapy plus
hyperthermia to metastatic lymphnodes in stage IV head
and neck patients. International Journal of Radiation Oncology, Biology, and Physics, 28:163169.
84
34. Dahl O. Interaction of heat and drugs in vitro and in vivo. In:
Seegenschmiedt MH, Fessenden P, Vernon CC, editors. Thermoradiotherapy and Thermochemotherapy, Vol 1. Berlin:
Springer Verlag; 1995. pp 103121.
35. Westermann AM, Grosen EA, et al. A pilot study of whole body
hyperthermia and carboplatin in platinum-resistant ovarian
cancer. Eur J Cancer 2001;37:11111117.
36. Li DJ, Hou BS. A preliminary report on the treatment of
esophageal cancer by intraluminal microwave hyperthermia
and chemotherapy. Cancer Treatment Reports 1987;71: 1013
1019.
37. Sugemachi K, et al. Chemotherapy combined with or without
hyperthermia for patients with oesophageal carcinoma: a
prospective randomized trial. International Journal of
Hyperthermia 10:485493.
38. Kakehi M, et al. Multi-institutional clinical studies on
hyperthermia combined with radiotherapy or chemotherapy
in advanced cancer of deep-seated organs. International Journal of Hyperthermia 6:719740.
39. Falk RE, et al. Combination therapy for resectable and
unresectable adenocarcinoma of the pancres. Cancer 57:
685688.
40. Issels RD, et al. Ifosfamide plus etoposide combined with
regional hyperthermia in patients with locally advanced sarcomas: a phase II study. Journal of Clinical Oncology
1990;8:18181829.
41. Issels RD, et al. Improvement of local control by regional
hyperthermia combined with systemic chemotherapy(ifosfamide plus etoposide) in advanced sarcomas: updated report on
65 patients. Journal of Cancer Research and Clinical Oncology
1991;117:141147.
42. Issels RD, et al. Preoperative systemic etoposide/ifosfamide/
doxorubicin chemotherapy combined with regional hyperthermia in high-risk sarcoma: a pilot study. Cancer Chemotherapy
and Pharmacology 1993;31(suppl. 2) S233S237.
43. Issels RD, et al. Neoadjuvant chemotherapy combined
with regional hyperthermia (RHT) followed by surgery
and radiation in primary and recurrent high-risk soft-tissue
sarcomas (HR-STS) of adults (updated report). Journal of
Cancer Research and Clinical Oncology 1998;124(suppl.)
R105.
44. Issels RD. Soft Tissues Sarcomas-What is Currently Being
Done. European Journal of Surgical Oncology 21(suppl) 471
474.
45. Eggermont A MM, et al. Isolated Limb Perfusion with highDose tumor necrosis factor alpha in combination with interferon gamma and melphalan for nonresectable extremity soft
tissue sarcomas: a multicenter trial. Journal of Clinical Oncology 14:26532665.
46. Wiedemann GJ, et al. Ifosfamide and carboplatin combined
with 41.8 8C whole body hyperthermia in patients with refractory sarcoma and malignant teratoma. Cancer Research
54:53465350.
47. Weidemann GJ, et al. Ifosfamide carboplantin, and etopside
(ICE) combined with 41.8 8C whole-body hyperthermia in
patients with refractory sarcoma. European Journal of Cancer
32A:888-891.
48. Robins HI, et al. Phase I Clinical Trail of Melphalan and
41.8 8C whole-body hyperthermia in cancer patients. Journal
of Clinical Oncology 1993;15:154164.
49. Romanowski R, Schott C. Regionale hyperthermie mit
systemischer Chemotherapie bei Kindern und Jugendlichen:
Durchfuhrbarkeit und Klinische Verlaufe bei 34 intensiv vorbehandelten Patienten mit prognostisch ungunstigen Tumorerkrankkungen. Klinische Padiatrie 205:249256.
50. Wessalowski R, Kruck H. Hyperthermia for the treatment of
patients with malignant germ cell tumors. A phase I/II study
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
in ten children and adolescents with recurrent or refractory

tumors. Cancer 82:793800.
Rietbroek RC, Schilthuis MS. Phase II trial of weekly locoregional hyperthermia and cisplantin in patients with previously irridated recurrent carcinoma of the uterine cervix.
Cancer 1997;79:935942.
Takahashi M, Fujimot S. Clinical outcome of intraoperative
pelvic hyperthermochemotherapy for patients with Dukes C
rectal cancer. International Journal of Hyperthermia 10:749
754.
Westermann AM, Wiedemann GJ. A Systemic Hyperthermia
Oncologic Working Group trial. Ifosfamide, carboplatin, and
etoposide combined with 41.8 degrees C whole-body
hyperthermia for metastatic soft tissue sarcoma. Oncology.
2003;64(4):31221.
Anhalt DP, Hynynen K, Roemer RB. Patterns of changes
of tumour temperatures during clinical hyperthermia:
Implications for treatment planning, evaluation and control.
Int J Hyperthermia 1995;11:425436.
Corry PM, Barlogie B, Tilchen EJ, Armour EP. Ultrasound
induced hyperthermia or the treatment of human superficial
tumors. Int J Rad Oncol Biol Phys 1982;8:12251229.
Corry PM, et al. Combined ultrasound and radiation therapy
treatment of human superficial tumors. Radiology 1982b;145:
165169.
Harrison GH. Ultrasound hyperthermia applicators: Intensity
distributions and quality assurances. Int J Hyperthermia
1990;6:169174.
Marmor JB, Hahn GM. Ultrasound heating in previously
irradiated sites. Int J Rad Oncol Biol Phys 1978;4:1029
1032.
Marmor JB, Hahn GM. Combined radiation and hyperthermia
in superficial human tumors. Cancer 1980;46:19861991.
Marmor JB, Pounds D, Hahn GM. Clinical studies with ultrasound induced hyperthermia. Natl Cancer Inst Monogram
1982;61:333337.
Marmor JB, Pounds D, Hahn GM. Treatment of superficial
human neoplasms by local hyperthermia induced by ultrasound. Cancer 1979;43:188197.
Labthermics Technologies Online.19981999. Available at
http://www.labthermics.com/hyper.html. Accessed 2005 Jan
26.
Anhalt DP, et al. Scanned ultrasound hyperthermia for treating superficial disease. Hyperthermic oncology. vol 2. Proceedings of the 6th international Congress on Hyperthermic
Oncology, Tucson, (AZ); 1992. p 191-192.
Lele PP. Advanced ultrasonic techniques for local tumor
hyperthermia. Radiol Clin N Am 1989;27:559575.
Harari PM, et al. Development of scanned focused ultrasound
hyperthermia: clinical response evaluation. Int J Rad Oncol
Biol Phys 1991;21:831840.
Hand JW, Vernon CC, Prior MV. Early experience of a commercial scanned focused ultrasound hyperthermia system. Int
J Hyperthermia 1992;8:587607.
Guthkelch AN, et al. Treatment of malignant brain tumors
with focused ultrasound hyperthermia and radiation: Results
of a phase I trial. J Neuro Oncol 1991;10:271284.
Duthkelch et al.
Formine et al.
Straube WL, et al. An ultrasound system for simultaneous
ultrasound hyperthermia and photon beam irradiation. Int J
Rad Oncol Biol Phys 1996;36:11891200.
Lu XQ, et al. Design of an ultrasonic therapy system for breast
cancer treatment. Int J Hyperthermia 1996;12:375399.
Moros EG, Fan X, Straube WL. An investigation of penetration
depth control using parallel opposed ultrasound hyperthermia. J Accoust Soc Am 1997;101:17341741.
73. Moros EG, Fan X, Straube WL, Myerson RJ. Numerical and
in vitro evaluation of temperature fluctuations during
reflected-scanned planar ultrasound hyperthermia. Int J
Hyperthermia 1998;14:367382.
74. Moros EG, Myerson RJ, Straube WL. Aperture size to therapeutic volume relation for a multi-element ultrasound system: Determination of applicator adequacy for superficial
hyperthermia. Med Phys 1993;20:13991409.
75. Moros EG, Roemer RB, Hynynen K. Simulations of scanned
focused ultrasound hyperthermia. The effects of scanning
speed and pattern on the temperature fluctuations at focal
depth. IEEE Trans Ultrason Ferroelec Frequency Control
1988;35:552560.
76. Moros EG, et al. Simultaneous delivery of electronic beam
therapy and ultrasound hyperthermia using scanning reflectors: a feasibility study. Int J Rad Oncol Biol Phys 1995;31: 893
904.
77. Moros EG, Straube WL, Myerson RJ. Potential for power
deposition conformability using reflected-scanned planar
ultrasound. Int J Hyperthermia 1996;12:723736.
78. Lele PP, Parker KJ. Temperature distributions in tissues
during local hyperthermia by stationary or steered beams of
unfocused or focused ultrasound. Br J Cancer 1982;45
(Suppl):108121.
79. Hynynen K, et al. A scanned, focused, multiple transducer
ultrasonic system for localized hyperthermia treatments. Int
J Hyperthermia 1987;3:2135.
80. Lin W, Roemer RB, Hynynen K. Theoretical and experimental evaluation of a temperature controller for scanned focused
ultrasound hyperthermia. Med Phys 1990;17:615625.
81. Ibbini MS, Cain CA. The concentric-ring array for ultrasound
hyperthermia: combined mechanical and electrical scanning.
Int J Hyperthermia 1990;6:401419.
82. Umemura S, Cain CA. The sector-vortex phased array: acoustic field synthesis for hyperthermia. IEEE Trans Ultraon
Ferroelec Frequency Control 1989;36:249257.
83. Ebbini ES, Cain CA. Experimental evaluation of a prototype
cylindrical section ultrasound hyperthermia phased array
applicator. IEEE Trans Ultrason Ferroelec Frequency Control 1991a;38:510520.
84. Ebbini ES, Cain CA. A spherical-section ultrasound phased
array applicator for deep localized hyperthermia. IEEE
Trans Biomed Eng 1991b;38:634643.
85. Benkeser PJ, Frizzell LA, Goss SA, Cain CA. Analysis of a
multielement ultrasound hyperthermia applicator. IEEE
Trans Ultrason Ferroelec Frequency Control 1989;36:319
325.
86. Diederich CJ, Hynynen K. Induction of hyperthermia using
an intracavitary ultrasonic applicator. IEEE Ultrason Sympos Proc 1987;2:871874.
87. Diederich CJ, Hynynen K. Induction of hyperthermia using
an intracavitary multielement ultrasonic applicator. IEEE
Trans Biomed Eng 1989;36:432438.
88. Diederich CJ, Hynynen K. The development of intracavitary
ultrasonic applicators for hyperthermia: A design and experimental study. Med Phys 1990;17:626634.
89. Smith NB, Buchanan MT, Hynynen K. Transrectal ultrasound applicator for prostate heating monitored using MRI
thermometry. Int J Rada Oncl Biol Phys 1999;33:217225.
90. Diederich CJ. Ultrasound applicators with integrated cathetercooling for interstitial hyperthermia: Theory and preliminary
experiments. Int J Hyperthermia 1996;12:279297.
91. Diederich CJ, Hynynen K. Ultrasound technology for interstitial hyperthermia. In: Seegenschmiedt MH, Sauer R, editors. Interstitial and intracavitary thermoradiotherapy. Berlin:
Springer-Verlag; 1993. pp 5561.
85
92. Hynynen K. The feasibility of interstitial ultrasound

hyperthermia. Med Phys 1992;19:979987.
93. Hynynen K, Davis KL. Small cylindrical ultrasound sources
for induction of hyperthermia via body cavities or interstitial
implants. Int J Hyperthermia 1993;9:263274.
94. Lee RJ, Klein LJ, Hynynen K. A multi-element and multicatheter ultrasound system for interstitial hyperthermia.
IEEE Trans Biomed Eng 1999.
95. Deardorff DL, Diederich CJ, Nau WH. Air-cooling of directcoupled ultrasound for interstitial hyperthermia and thermal
coagulation. Med Phys 1998;25:24002409.
96. Diederich CJ, et al. Direct coupled interstitial ultrasound
applicators for simultaneous thermobrachytherapy: A feasibility study. Int J Hyperthermia 1996;12:401419.
97. Jarosz BJ. Feasibility of ultrasound hyperthermia with
waveguide interstitial applicator. IEEE Trans Biomed Eng
1996;6:11061115.
98. Kennedy J, et al. High intensity ultrasound: Surgery of the
future? Br J Rad 2003;76:590599.
99. Lynn JG, Zwemer RL, Chick AJ, Miller AG. A new method for
generation and use of focused ultrasound in experimental
biology. J Gen Physiol 1942;26:179193.
100. Fry WJ, et al. Ultrasonic lesions in the mammalian central
nervous system. Science 1955;122:517518.
101. Fry WJ, Mosberg WH, Barnard JW, Fry FJ. Production of
focal destructive lesions in the central nervous system with
ultrasound. J Neurosurg 1954;11:471478.
102. Fry FJ. Precision high-intensity focusing ultrasonic
machines for surgery. Am J Phys Med 1958;37:152156.
103. Ballantine HT, Bell E, Manlapaz J. Progress and problems in
the neurological application of focused ultrasound. J Neurosurg. 1960;17:858876.
104. Warwick R, Pond JB. Trackless lesions in nervous tissues
produced by HIFU (high-intensity mechanical waves). J Anat
1968;102:387405.
105. Lele PP. Concurrent detection of the production of ultrasonic
lesions. Med Biol Eng 1966;4;451456.
106. Lele PP. Production of deep focal lesions by focused ultrasound-current status. Ultrasonics 1967;5:105112.
107. Burov AK. High-intensity ultrasonic vibrations for action on
animal and human malignant tumours. Dokl Akad Nauk
SSSR 1956;106:239241.
108. Kohrmann KU, et al. Technical characterization of an ultrasound source for noninvasive thermoablation by high-intensity focused ultrasound. BJU Int 2002;90:248252.
109. ter Haar G. Robertson D: Tissue destruction with focused
ultrasound in vivo. Eur Urol 1993;23(Suppl. 1):811.
110. Clement GT, Perspectives in clinical uses of high-intensity
focused ultrasound. Ultrasonics 2004;42:10871093.
111. Holt RG, Roy RA, Edson PA, Yang X. Bubbles and HIFU: the
good, the bad, and the ugly. In: Andrew MA, Crum LA, Vaezy
S, editors. Proceedings of the 2nd International Symposium
on Therapeutic Ultrasound; 2002. pp 120131.
112. Sokka SD, King R, Hynynen K. MRI-guided gas bubble
enhanced ultrasound heating in in vivo rabbit thigh. Phys
Med Biol 2003;48:223241.
113. Billard BE, Hynynen K, Roemer RB. Effects of physical
parameters on high temperature ultrasound hyperthermia.
Ultrasound Med Biol 1990;16:409420.
114. Kolios MC, Sherar MD, Hunt JW. Blood Flow cooling and
Ultrasonic lesion formation. Med Phys 1996;23:12871298.
115. Chan A, et al. An image-guided high intensity focused ultrasound device for uterine fibroids treatment. Med Phys
2002;29:26112620. Otsuka R, et al. In vitro ablation of
cardiac valves using high intensity focused ultrasound.
86
HYPOTHERMIA.
116. Wu F, et al. Pathological changes in human malignant

carinoma treated with high-intensity focused ultrasound.
117. Chaussy C, Thuroff S. High-intensity focused ultrasound in
prostate cancer: Results after 3 years. Mol Urol 2000;4:179
182.
118. Madersbacher S, et al. Tissue ablation in benign hyperplasia
with high-intensity focused ultrasound. Eur Urol 1993;23
(Suppl. 1):3943.
119. Hynynen K, et al. MR imaging-guiding focused ultrasound
surgery of fibroadenomas in the breast: A feasibility study.
Radiology 2001;219:176185.
120. Hynynen K, et al. A clinical noninvasive MRI monitored
ultrasound surgery method. RadioGraphics 1996;16:185195.
121. Hazel JD, Stafford RJ, Price RE. Magnetic resonance
imaging-guided focused ultrasound thermal therapy in
experimental animal models: Correlation of ablation volumes
with pathology in rabbit muscle and VX2 tumors. J Magnet
Reson Imag 2002;15:185194.
122. Weidensteiner C, et al. Real time MR temperature mapping
of rabbit liver in vivo during thermal ablation. Mag Reson
Imag 2003;50:322330.
123. Chan AH, et al. An Image-guided high intensity focused
ultrasound device for uterine fibroids treatment. Med Phys
2002;29:26112620.
124. Tempany CMC, et al. MR imaging-guided focused ultrasound
surgery of uterine leiomyomas: A feasibility study. Radiology
2003;226:897905.
125. Stewart EA, et al. Focused Ultrasound treatment of uterine
fibroid tumors: Safety and feasibility of a noninvasive thermoablative technique. Am J Obstet Gynecol 2003;189:4854.
126. Wu F, et al. Changes in ultrasonic image of tissue damaged
by high intensity ultrasound in vivo. J acoustic Soc Am
1998;103:2869.
127. Wu F, Wang ZB, Chen WZ, Zou JZ. Extracorporeal HighIntensity Focused Ultrasound for treatment of solid carcinomas: Four-year Chinese clinical experience. Proceedings of
the 2nd International Symposium on Therapeutic Ultrasound; July 29-Aug1; Seattle; 2002.
128. Kennedy JE, et al. High- Intensity focused ultrasound for the
treatment of liver tumours. Ultrasonics 2004;42:931935.
129. Wu F, et al. Extracorporeal high intensity focused ultrasound
ablation in the treatment of patients with large hepatocellular carcinoma. Ann Surg Oncol 2004;11(12): 10611069.
130. Uchida T, et al. Transrectal high-intensity focused ultrasound for treatment of patients with stage T1b-2NOMO
localized prostate cancer: A preliminary report. Urology
2002;59:394399.
131. Blana A, Walter B, Rogenhofer S, and Wieland W. HighIntensity focused ultrasound for the treatment of localized
prostate cancer: 5-year experence. Urology 2004;63:297300.
132. Madersbacher S, et al. Effect of high intensity focused ultrasound on human prostate cancer in vivo. Cancer Res
1995;55:33463351.
133. Gelet A, et al. Treatment of prostate cancer with transrectal
focused ultrasound: Early clinical experience. Eur Urol
1996;29:174183.
134. Chaussy C, Thuroff S, Lacoste F, Gelet A. HIFU and prostate
cancer: The European experience. Proceedings of the 2nd
International Symposium on Therapeutic Ultrasound; July
29Aug 1; Seattle; 2002.
135. Beerlage HP, et al. High-intensity focused ultrasound
followed after one to two weeks by radical retropubic prostatectomy: Results of a prospective study. Prostate 1999;39:41
46.
136. Beerlage HP, et al. Transrectal high-intensity focused ultrasound using the Ablatherm device in treatment of localised
prostate carcinoma. Urology 1999;54:273277.
137. Kohrmann KU, et al. High-intensity focused ultrasound for
noninvasive tissue ablation in the kidney, prostate, and
uterus. J Urol 2000;163 (4Suppl.):156.
138. Wu F, et al. A randomized clinical trial of high-intensity
focused ultrasound ablation for the treatment of patients
with localised breast cancer. BJC 2003;89:22272233.
139. Mc Pherson K, Steel CM, Dixon JM. Breast cancer epidemiology, risk factors, and genetics. Br J Med 2000;321:
624628.
140. Curran D, et al. Quality of life in early stage breast
cancer patients treated with radical mastectomy or breastconverging procedure: Results of EORTC trial 10801. The
European Organization for Research and Treatment of
Cancer (EORTC), Breast Cancer Cooperative Group (BCCG).
Eur J Cancer 1998;34:307314.
141. Aubry J, et al. Experimental demonstration of noninvasive
transskull adaptive focused based on prior computed tomography scans. J Acoust Soc Am 2003;113:8493.
142. Clement GT, Hynynen K. A noninvasive method for focusing
ultrasound through the human skull. Phys Med Biol
2002;47:12191236.
143. Foley J, et al. Image-guided HIFU neurolysis of peripheral
nerves to treat spasticity and pain. Ultrasound Med Bio
2004;30:11991207.
144. Ohtsuka T, et al. Thoracoscopic internal mammary artery
harvest for MICABG using the Harmonic Scalpel. Ann
Thorac Surg 1997 June; 63 (6Suppl):S10.
145. Lee LA, et al. High intensity focused ultrasound effect on
cardiac tissues: Potential for clinical application. Echocardiography 2000 Aug; 17(6 Pt 1):563566.
146. Strickberger SA, et al. Extracardiac ablation of the canine
atrioventricular junction by use of high-intensity focused
ultrasound. Circulation 1999;100:203208.
147. Natale A. Cleveland Clinic Foundation. Personal Interview.
March 5, 2005.
148. Natale A, et al. First Human Experience with Pulmonary
vein isolation using a through-the-balloon circumferential
ultrasound ablation system for recurrent atrial fibrillation.
Circulation 2000;102:1879.
149. Saliba W, et al. Circumferential ultrasound ablation for pulmonary vein isolation: Analysis of acute and chronic failures. J of
Cardiovasc Electrophysiol 2002;13:957961.
150. Ayoma H, et al. Circumferential lesion characteristic of high
intensity focused ultrasound balloon catheter for pulmonary
vein isolation. Heart Rhythm 2004;1:S430.
151. Smith NB, Hynyen K. The feasibility of using focused ultrasound for transmyocardial revascularization Ultrasound
Med Biol 1998;24:10451054.
152. Foldes K, et al. Magnetic resonance imaging-guided focused
ultrasound synovectomy. Scand J Rheumal 1999;28:233237.
153. Vaezy S, et al. Hemostasis of punctured blood vessels using
high-intensity focused ultrasound. Ultrasound Med Biol
1998;24:903910.
154. Roberts WW, et al. High-Intensity focused ultrasound
ablation of the vas deferens in canine model. J Urol 2002;
167:26132617.
See also HYPERTHERMIA,
INTERSTITIAL; HYPERTHERMIA, SYSTEMIC;
THERMOMETRY.
HYPOTHERMIA. See TEMPERATURE MONITORING.
I
IABP. See INTRAAORTIC BALLOON PUMP.
obviated and fine details could be seen. An optical viewer

comprising an objective lens and relay lens, similar in
design to a submarine periscope, was used to observe
the image.
Modern image intensifiers use an epitaxially grown
scintillator of CsI with the photoemitter deposited directly
on the surface of that layer. Because the epitaxial layer can
be made much thicker than the powdery deposit of the
older type scintillator for the same resolution, the new
image intensifiers require less exposureimage than conventional radiographs. The thicker layer has a higher QDE
than the fluoroscopic screens (Fig. 1).
Many modern image intensifiers use a thin curved steel
window on which the scintillator and photoemitter are
deposited. This geometry eliminates the X-ray scatter
produced by the glass window of the older tubes and
improves image contrast. Larger tubes up to 35 cm sensor
diameter have been made with variable magnification or
zoom capability. The window of thin steel is the flash with a
coating of aluminum or other metal and etched to create a
domain structure, similar to the domains seen in the zinc
coating of galvanized steel. The domains here are very
small, 100 mm diameter. Cesium iodide is then vapor
deposited on this surface and forms epitaxially (i.e., crystal
growth follows the orientation of the metallic substrate),
and grows as a collection of optically isolated fibers. This
scintillator can be made quite thick with little light spreading. The greater thickness means higher quantum efficiency (i.e., a measure of the fraction of incident X-ray
photons converted to light photons), when compared to
conventional phosphor plate scintillators. A thin layer of
silver and antimony is vapor deposited on top of the scintillator and the final sensitization is accomplished by
depositing cesium from heated tubes or reservoirs after
assembly in the vacuum tube. The AgCs:Sb photoemitter
on the surface of the scintillator is similar to the photocathode of a photomultiplier (PMT), tube.

MELVIN P. SIEDBAND
University of Wisconsin
Fitchburg, Wisconsin
INTRODUCTION
Early fluoroscopic systems used a phosphor-coated sheet or
screen to convert incident X-ray photons to light. The
radiologist observed the image through a lead glass protective screen. A film camera was often used to record the
image. There are two serious disadvantages to this method:
the radiologist had to be dark adapted so that image details
were hard to see and the collection solid angle of the eye or
camera lens was small. The small collection angle meant
that the X-ray exposure had to be increased by almost 100
times to have the same diagnostic quality as a conventional
radiograph. Photographing the fluoroscopic screen, photofluorography, is no longer used because of the high exposure to the patients. All X-ray images are noise limited by
the finite number of X-ray quanta detected and seen. A 50
keV X-ray photon can, at best, produce 1000 visible light
photons, if absorbed by the old-type phosphor. About 5
10% of the incident X-ray photons are stopped and converted to light by the fluorescent screen. The quantum
detection efficiency, (QDE) of the screen is the product of
the ability to absorb the incident X-ray photon and the
probability of emitting light. A thicker screen would absorb
more photons, but would also cause more lateral spreading
of the light and reduce the resolution of the image. The 5
10% QDE figure is a practical compromise between resolution and sensitivity. If it is assumed that the visible light
photons are emitted isotropically, then the lens of the eye
or camera subtends only a very small fraction of this light
radiation hemisphere. A sheet of film in a radiographic
cassette has a phosphor-coated sheet on either side and can
collect light photons far more efficiently.
The invention of the image intensifier overcame these
objections. The concept of the early image intensifiers was
to use a thin, curved, glass meniscus, 15 cm diameter,
coated on the convex side with a scintillator, originally of
the same composition as the zinc:cadmium sulfide fluoroscopic phosphor plates, and a photoemitter on the concave
surface. Light produced by the scintillator did not have far
to travel to excite the photoemitter. This assembly was
placed in a vacuum tube and the photoelectrons were
accelerated toward an output viewing screen where a small
and very bright image was produced. Because the photoemitter was in close optical contact with the scintillator,
the collection angle was very large so that a higher radiation exposure was not needed and the image at the viewing
screen was bright enough so that dark adaptation was
Figure 1. In the image intensifier, X rays strike the input

phosphor screen, thus generating light. Light stimulates the
photocathode to emit electrons, which are accelerated through
25 kV to strike the output phosphor screen. Brightness gain is
due to both geometric gain and electronic gain.
87
88
The vacuum tube uses a series of metal cylinders between

the photoemitter and the output phosphor. The photoemitter, cylinders, and output phosphor are connected to voltage
sources to create shaped electric fields. The field potential at
any point affects an electron beam in the same way that the
index of refraction of a glass lens affects a beam of light.
These cylinders and their potentials form electrostatic
lenses to focus the photoelectrons to produce the small
and bright image on the output phosphor. Like optical lens
elements, the metal cylinders can make compound electrostatic lenses so that focal length can be varied to change the
size of the output image: variable zoom.
The brightness gain of an image intensifier is a measure
that compares the brightness of the image at the small
output screen to that of the older fluoroscopic screens. This
gain is a result of the added energy imparted to the
photoelectrons, the higher probability of stopping an incident X-ray photon, and the effect of compressing the large
input area signal to the small area of the output screen. The
older term, brightness gain, has been replaced by conversion efficiency. This term is defined as the output luminance in candelam2 for an input exposure rate of
1 mRs1 or 10 mGys1. If assumed it is that a single
50 keV input X-ray photon has a 50% probability of interacting with the input scintillator and producing light (i.e.,
QDE is 50%) and produces 2000 visible light photons, of
which 1000 reach the photoemitter. About 100 of these
will produce electrons that will be accelerated by the
electric field across the tube and strike the output phosphor
as 25 keV electrons. Each 25 keV photoelectron has about
a 10% probability of converting its energy into 2.2 eV
visible light photons. The QDE, quantum detection efficiency, of the image intensifier is assumed to be 50%
because of the thicker CsI scintillator.
An input exposure rate in the diagnostic range produces
200,000 X-ray photonsmm2s1, converts to 100,000
light photons/X-ray photon. In the visible range, 1010
photons/mm2 have a light intensity of 1 candela/m2. Substituting in the above, a 1 mR (10 mGy) input would produce 1 candelam2 for an output screen the same size as
the input scintillator. This is >200 times brighter than the
older fluoroscopic screens. Because the output screen area
of an image intensifier is 25 mm diameter for an input
screen of diameter of 220 mm, the same number of output
light photons are emitted from a smaller area resulting in
an additional gain factor of 75. This yields a total brightness gain of >15,000 over the old fluoroscopic screen!
A fluoroscopic television system is used for dynamic
studies, to enable the viewer to see how a contrast agent
is swallowed, how blood flows, to locate objects for a surgical procedure, and so on. During the study, an image record
(formerly a film record) could be made for later examination using a video recorder or computer. A rapid succession
of images could be recorded and then viewed to find the few
images revealing the particular problem, for example, how
a heart valve functioned or to diagnose an eroded region in
the esophagus as the contrast medium sped by. Or a single
image could be made to show the problem for later correction.
A fast optical lens optically collimates the small output
image of the image intensifier. Collimation, in this case,
means that the optical object is at the focal plane of the

collimating lens and its image is focused on an infinite
distance away. Any optical device with its own objective
lens, such as a TV or film camera, could be aimed through
the collimating lens and its image size would be the original
intensified image size times the ratio of (objective focal
length)/(collimator focal length). The lenses could be separated by several centimeters before image vignetting
occurs. A beam splitting mirror can be interposed on the
collimator and the objective lenses so that image light is
simultaneously apportioned between a TV and film camera. A small mirror or prism and lens could sample the light
and form an image over a small hole in the cover of a PMT
tube. The hole would permit only light from the center of
the image to reach the PMT. The output of the PMT is used
to control the X-ray tube current to maintain constant
image brightness for continuous viewing or for automatic
exposure control of one or a sequence of recorded images.
This automatic brightness stabilizer scheme is similar
in concept to automatic exposure control of most digital
cameras.
The quality of an X-ray image is a compromise between
exposure to the patient and the noise or graininess of the
image. Most images are made using the ALARA principal
(As Low As Reasonably Achievable). For any given X-ray
exposure, there are a finite number of X-ray photons
incident on the patient and then, through the patient,
incident on the image sensor. The statistics of photons
area follow the Poisson distribution, so that the variance
(noise) is the square root of the average number of photons
in a pixel (picture element). To produce a second image
having twice the linear resolution (detail) as the first
requires four times the exposure. Because the eye averages
exposure time >0.2 s, to record an equivalent image in 0.02
s requires an exposure rate 10 times greater. X-ray exposure requirements are determined by the by the X-ray
absorption of the patient, diagnostic needs and are different for continuous viewing (real-time fluoroscopy), a
sequence of images (video or motion pictures), or single
images for later diagnoses.
Before the rapid growth and improvement of digital
cameras and computer technologies, still and motion picture film cameras were the only practical means to record
images. Because of differences of integrating capability of
the eye and detail required of the recorded film images, the
X-ray beam current, pulse width (exposure time/image),
and so on, the ratio of transmission/reflection of the beam
splitting mirror, and other operating parameters, must, be
adjusted to obtain the required image quality for each
image application requirement while minimizing total
exposure to the patient.
Most modern systems use charge-coupled image sensors
with a high dynamic range and pulse the X-ray beam
current to the required level while digitally recording video
images. Computer display of selected images or a dynamic
sequence of images has largely displaced motion picture
film techniques. Video tape recorders are used in simple
systems. The automatic brightness controlautomatic
exposure control of the digital system uses signals from
selected image areas derived from the computer image to
optimize image quality in the region of interest.
IMAGING DEVICES
BIBLIOGRAPHY
Further Reading
Gebauer A, Lissner J, Schott O. Roentgen Television. New York:
Grune & Stratton; 1966.
Siedband MP. Image storage subtraction techniques and contrast
enhancement by electronic means. Symposium on the Physics
of Diagnostic Radiology; University of California, June 1968.
Siedband MP. Image intensification and television. In: Taveras,
Ferrucci, editors. Radiology, Diagnosis, Imaging, Intervention.
Chapt. 10. New York: Lippincott; 1990.
Siedband MP, Duffy PA. Brightness Stabilizer with Improved
Image Quality, US patent No. 3,585,391. Accessed 1971.
Siedband MP. X-ray image storage, reproduction and comparison
system. US patent No. 3,582,651, 1971.
IMAGING, CELLULAR. See CELLULAR IMAGING.
IMAGING DEVICES
MARK J. RIVARD
Tufts New England Medical
Center
FRANK VAN DEN HEUVAL
Wayne State University
INTRODUCTION
Historically, external radiation treatment of deep-seated
malignancies was performed using ortho-voltage equipment. The radiological characteristics of these beams
caused maximum dose deposition to occur on the skin of
the patient. At that time, skin damage was the limiting
factor for dose delivery to the tumor. When the skin turned
red due to radiation damage (erythema), the physician had
to find another area or portal through which to deliver
radiation. The portal was then defined by its orientation
and the surface of skin it irradiated.
Nowadays, the treatment is performed with higher
photon energies that permit a skin-sparing effect (i.e.,
the dose at the skin is lower than that deposited a few
centimeters deeper) due to the absence of electronic equilibrium. The historic name portal still denotes a radiotherapy treatment beam oriented for entry within a
patient. Physicians verify whether the treatment is correct
using megavoltage treatment beams (420 MV photons) as
an imaging tool. A transmission image, obtained much like
a diagnostic transmission image, provides information
describing the patient anatomy and gives clues on the
beam orientation and positioning, but also on the extent
and shape of the treated area (or portal field). As such,
portal imaging is the most direct manner to confirm accuracy of treatment delivery.
Traditional portal verification is done using radiographic films, much like the classical diagnostic films.
Films are positioned at the beam exit side of the irradiated
patient. Portal image analysis involves comparison with a
simulation image that is typically obtained using diagnos-
89
tic quality X rays (60120 kV photons). The simulation

image serves as the reference image, showing anatomical
information clearly and delineating the intended treatment field. Comparison of the simulation image with the
portal image is complicated due to the inherent poor quality obtained when imaging using high energy photons (1).
The whole procedure of patient positioning, artifact removing, imaging processing, and evaluation using film represents a significant fraction of the total treatment time. This
procedure increases the workload per patient, and as a
result, the number of images taken is minimized due to
economic concerns rather than concerns for efficiency or
treatment quality. Indeed, studies have demonstrated that
weekly portal image verification, which is the current
clinical standard, does not guarantee accurate treatment
setup for a population of patients (2).
Portal imaging differs considerably from diagnostic
transmission imaging. The main difference is the photon
energies used to generate the images. In diagnostic imaging, photons having energies ranging from 50 to 120 kV
interact in patients primarily via the photoelectric effect.
The cross-section for these interactions is highly dependent
on the atomic number of the medium in which they traverse: A higher atomic number increases the probability of
interaction. The average atomic number of bony anatomy
is higher than that of soft-tissue, yielding good contrast for
the bony anatomy. At treatment energies (110 MeV) the
predominant photon interaction is Compton scattering.
The cross-section for this interaction is largely dependent
on the media density, and the resulting image will show the
largest contrast when large differences in density are
present. In practice, this means that differences in soft
tissues will contribute most to the visible signal.
These considerations imply that the dynamic range of
an electronic portal imaging detector (EPID) is used to
obtain information on soft-tissue variations (3), divergence
effects (4), scatter contributions (5), field-edge information,
and in the case of fluoroscopic imagers: vignetting and
glare. With the exception of field-edge information, all of
these factors are nonlocalized and tend to change gradually
within an image. Not only are these features slowly varying, but they also have a large contrast-to-noise ratio (CNR)
compared to the clinically important bonesoft-tissue contrast.
The EPIDs permit the same tasks as film-based imaging, but increase the efficiency and provide added value by
using digital imaging techniques. The EPIDs are devices
that electronically capture the photon energy fluence
transmitted through a patient irradiated during treatment, and allow direct digitization of the image. This image
is then immediately available for visualization on a computer screen and electronic storage. When the treatment
verification process uses EPIDs, departmental efficiency is
increased and quality is improved at the same cost as when
using film-based imaging.
Proposals to generate electronic images started in the
beginning of the 1980s mainly through the work of Bailey
et al. (6), who used systems based on video techniques. This
seminal work was then further developed toward more
clinically applicable systems by Shalev and co-workers
(7), Leong (8), Munro et al. (9), and Visser et al. (10). All
90
IMAGING DEVICES
of these systems were the basis for the first generation of

commercially available EPIDs. They all combined an analog camera with a fluorescent screen generating the optical
coupling using a mirror system. Wong et al. (11) replaced
the mirror system with optical fibers (one for each pixel).
The technology developed further, and is described below
in greater detail.
PHYSICAL ASPECTS OF ELECTRONIC PORTAL IMAGING
TECHNOLOGY
Camera-Based Detectors
The initial experience using EPIDs was obtained using
camera-based systems. Again similar to film-based portal
imaging, the camera-based systems measured the photon
energy fluence exiting the patient. However, phosphorescent and fluorescent screens replaced the film, and a mirror
was oriented at an angle of 458 to reflect the screen toward
a video camera. Subsequently, the image was digitized.
Because of the intrinsically low-detector efficiency, bulky
detector size, and poor image quality, this technology has
now become outdated in comparison with more sophisticated technologies.
The low-detector efficiency was due to many limitations
in the signal path from screen to computer. Screen conversion efficiency was not ideal when using Gd2O2S. In
addition, <0.1% of the light emitted reached the video
camera, due to the poor light collection efficiency of the
video camera lens. This low rate of signal collection was
subsequently impacted by competing electronic noise from
the camera in close proximity to the operating linear
accelerator (linac). Also, image acquisition typically
required a full treatment fraction as compared to the
technique using partial fraction irradiation that is commonly used for radiographic portal imaging. Due to the
camera-based system detector orientation, rigid positioning of the large mirror was crucial. Changes in linac gantry
rotation could cause apparent changes in patient positioning due to physical sag of the camera and mirror mounting
system. Furthermore, image quality was suboptimal due to
the large lenses required to focus the light signal to the

video camera. Degradation of spatial resolution, field uniformity, signal-to-noise (SNR), and field flatness all contributed to minimizing the utility of this detector type.
LIQUID IONIZATION CHAMBERS
The liquid ionization chamber (LIC) is based on a design
proposed by Wickman (12), who proposed to use liquid as
an ionization medium to increase the efficiency of ionization chambers. Indeed, the introduction of isooctane
increased the signal level by over a factor of 10, but also
deleteriously increased the recombination of the electrons
due to their low mobility. A first prototype was built by
Meertens et al. (13) using two printed circuit boards with
perpendicular electrode strips. This resulted in a 30 30
matrix of ionization chambers, and was further refined by
van Herk et al. (14) to include 128 128 and finally
256 256 matrices.
To obtain an image, the matrix is scanned row by row, by
successively switching high voltage to different row electrodes and measuring all column electrodes. The ionization
chamber polarizing voltage is typically 300 V, which is
comparable to the voltage applied over a regular megavoltage ionization chamber. The typical current produced
by the chamber is of the order of 100 pA. Due to the high
voltage switching there is a limit on the speed with which
the image may be obtained.
In most of the commercially available imagers, 1 s is
required to readout the complete matrix. Figure 1 shows a
schematic diagram of a EPID. The LIC is efficient in that it
is able to obtain information constantly in between readouts. The low recombination rate (a 4.51016 m3/s) of the
ions in the liquid makes that the signal accumulates during
radiation and provides an averaging effect.
Multiple Detector Combinations
An alternative way to obtain two-dimensional (2D) transmission images is to use a line detector much like the ones
found in computertomography devices. They consist of a
1
2
3
HV
Chamber
electrodes
t
Electrode
voltages
To other
electrometers
Electrometer 1
I1,1
Figure 1. Implanted gold seeds, imaged using a
flat-panel portal imager.
I1,2 I1,3
t
Ionization current
IMAGING DEVICES
tages of these circuits is that they are highly radiation

resistant and can be placed directly in a radiation beam.
As with computer integrated circuit chip technology, the
TFT EPID can be etched with a resolution of a few
micrometers, permitting construction of a large detector
matrix. As the photodiodes only detect visible light, a
phosphorescent screen is used to perform the conversion
much as for camera-based EPIDs. The TFT EPID is nonconducting during the radiation. To read out the TFT, a
voltage bias is applied to allow collected charge to flow
between the photodiode and an external amplifier. An
amplifier records this charge, which is proportional to
the light intensity. The TFT EPID array has a maximum
readout rate of 25 Hz. In comparison to the camera-based
EPID system, the large TFT detectors are designed to
be in direct contact with the conversion screen, thus
eliminating the poor optical coupling and efficiency intrinsic to the camera-based systems.
line of point detectors, which usually contain a phosphorescent material and an optical light detector (15). Alternatively, Lam et al. (16) constructed a device containing
256 silicon diodes. The line of detectors is scanned through
the field in a mechanical fashion. However, this approach is
time intensive and not appropriate for clinical techniques
such as respiratory gated radiotherapy for treatment of
lung cancer where the exiting photon energy fluence is of a
dynamic nature.
Flat Panel Technology
The advance of flat-panel displays, where the use of
amorphous silicon created surfaces that locally behaved
as a crystalline material, allowed for lithography of integrated circuits. The same thin-film technology (TFT) was
used to generate photodiode circuits detecting optical
light. The TFT is deposited on a glass substrate of 1
mm thick as is shown in Fig. 2. One of the major advan-
Incident X ray
Indium tin oxide

p-doped Si
a-Si:H
n-doped Si
Metal
Dielectric
Converting phosphor
Passivation
Glass substrate
TFT
91
Photodiode
(a)
Figure 2. Schematic cross-section (not to scale) of a single a-Si:H imaging pixel.
92
IMAGING DEVICES
(a)
(b)
Figure 3. Illustration of EPID conversion screens used to trap optical light. Fig. 3a shows a regular
Gd2O2S-screen with optical spread, while Fig. 3b shows a CsI screen which limits optical spread and
increases detector resolution.
The resolution and efficiency of flat-panel imagers are

theoretically superior to those from camera-based and LIC
EPIDs. Research performed by Munro et al. (17) indicates
that the amorphous silicon imager is X-ray quantum limited, and that the resolution is limited by the spread of the
optical photons in the imager. The increase in quality and
the compact size of TFT EPIDs means that they may be
easily installed onto a linac using a robotic arm. Consequently, TFT EPIDs are now the only type of detectors
commercially available. Efforts are underway to replace
the current conversion screens, which usually are Gd2O2S
phosphors, with CsI, which can be grown as single crystals
the size of a pixel. The optical light is then trapped in a
manner similar to an optical fiber, and therefore optical
spread is eliminated, as shown in Fig. 3.
Figure 4 shows pelvic images taken with a camerabased detector, LIC, and flat-panel imager.
EPID APPLICATIONS
Replacement of Radiographic Film
Just as is common in radiology departments, electronic
acquisition and management of imaging data is quickly
becoming standard practice. Because of the fast pace of
detector evolution in the past decade, EPID technology is
facilitating hospital-wide imaging digitization. However,
to understand why widespread implementation of EPID
systems has not yet occurred, it is important to perform a
brief cost analysis.
Compared to radiographic film-based portal imaging,
up-front capital expenditures for EPID systems are
about a factor of 5 larger (e.g., $25,000 vs. 125,000).
However, on-going costs associated with an EPID system as compared to a radiographic film-based portal
imaging program are much less. For example, regular
purchasing of film and maintenance of a film processor
(including silver harvesting) is not required with an
EPID program. This cost-analysis makes EPID highly
competitive given the digital direction facing modern
health care.
With direct image digitization comes the ability to
transmit data as required for telemedicine. Furthermore,
imaging data storage and retrieval, as required for radi-
ology picture archiving and communication systems

(PACS), has additional advantages over conventional
radiographic film storage. In radiation oncology departments, it is now commonplace for record-and-verify systems to be coupled to an electronic patient charting system.
By storing the EPID images in this domain, many of the
concerns for patient record keeping, retrieval, and preservation of treatment confidentiality are overcome.
Improvement of Patient Positioning
The original purpose of portal imaging is to reduce the
incidence and extent of errors made during radiation treatment. The type of errors can be categorized as gross errors
and stochastic errors. Examples of such errors are shown in
Fig. 5. The gross errors occur only once and when detected
can be removed from the treatment after one fraction.
Stochastic errors contain a random component, which
implies that the error changes from day to day. Errors
and QA-problems can also introduce a systematic component hidden by the random component, which can only be
corrected for if its extent is known.
To reduce stochastic errors, there are two general methodologies, on- or off-line corrections. The most straightforward methodology uses on-line correction where an image
is obtained in localization mode where minimal dose to
the patient is applied. If a discrepancy is observed in the
patient setup, and if this discrepancy is larger than a
predetermined threshold or action level, efforts are taken
to eliminate the error by changing the patient position or
changing the treatment configuration. The aforementioned
threshold is based on the precision to which position can be
determined and with which the patient setup correction
can be applied. Given the digital nature of EPID images, it
is possible to increase the accuracy using computerized
algorithms to objectively measure patient positioning with
respect to the treatment field (1821). This approach has
not been widely adopted, mainly due to the perceived labor
intensity and some medico-legal aspects. However, this
may change with the advent of other on-line repositioning
techniques (cf. ultrasound-based repositioning) and
increased process automation.
Weekly port-filming is the standard procedure in the QA
of external beam radiation therapy, which generally
IMAGING DEVICES
93
Figure 4. A comparison of pelvic images taken with three different types of EPIDs: 4a) a camerabased system, 4b) a liquid ionization chamber system, and 4c) a amorphous silicon (a-Si:H) flat panel
imaging system.
implies that a sample of 4 positions is taken out of a 20

fraction treatment. Errors are corrected after the first
image. An interesting study by Valicente et al. (2), showed
that this practice is suboptimal. The use of EPIDs allows us
to obtain images in a more economical way. Most off-line
correction strategies assume that the distribution formed
by all consecutive errors is a normal distribution characterized by the mean error and the standard deviation
calculated as in
Mean:
hxi
N
1X
x
N i1 i
Standard Deviation:
v
u
N
u 1
X
hxi xi 2
st
N 1 i1
Figure 5. Examples of setup errors. Reference images on the right are digitally reconstructed
radiographs with the correct setup. On the left are the measured portal images. (a) Faulty collimator
angle. (b) Wrong blocking used. (c) Wrong MLC file. (d) Patient positioning error.
94
Figure 5. (Continued)
95
96
IMAGING DEVICES
By repeated sampling of the distribution (e.g., taking port

films), the strategy estimates the value of hxi as close as
possible and corrects for the error, which will reduce the
systematic error in the treatment. Several groups have
studied the implications of this strategy and its variations.
The most successful approach seems to be the following
strategy: The set-up variations are recorded and averaged.
This is compared to an action level that depends on how
many samples have been taken already (the level shrinks
as the the amount of information on the systematic error
increases). These studies showed that systematic errors
could be reduced to 2 mm (22).
Organ Motion
One of the major reasons for use of EPIDs is the fact that
the patients anatomy and position vary from those used for
treatment planning purposes. The factors involving this
variation are
Patient movement.
Patient positioning inaccuracies.
Organ motion.
Any of these factors will influence the actual dose distribution to be different from that obtained using treatment planning. It is straightforward to correct for the first
two problems using portal imaging as the patients position
is typically well-characterized using bony anatomy. An
excellent compilation on the incidence, extent, and repercussions of organ movement was performed by Langen and
Jones (23). Efforts to incorporate organ movement during
radiotherapy treatment planning involves enlarging the
target to be treated. Sophisticated algorithms that calculate the extent of these enlargements were developed
independently by Stroom et al. (24) and by van Herk
et al. (25). The general framework for this enlargement
is given in ICRU 50 and ICRU 60 (26,27). In these reports
the gross target volume (GTV) is defined as the volume
containing demonstrated tumor. A margin is added to the
GTV to account for suspected microscopic tumor involvement, and is defined as the clinical target volume (CTV).
Finally, the planning target volume (PTV) is defined by the
CTV and an additional margin to allow for geometrical
variations such as patient movement, positioning errors,
and organ motion. The margins added to GTV and CTV can
substantially increase the PTV since the margins are
applied in three dimensions. Because of volume effects of
radiation therapy, there is a tendency to minimize the PTV
by increasing the precision of the treatment. As explained
above, EPIDs are able to minimize uncertainties caused by
patient motion and positioning errors using off- or on-line
correction strategies.
Except in a few cases like in lung or where air pockets
are present (24,28), the target is virtually indiscernible
using X rays. To solve this problem, other modalities, like
CT (29) and ultrasound (30,31), have been used to determine the position of the organ. The EPIDs can also be
used to image the position of organs if radioopaque
markers are implanted. The markers need to be of high
density and migration needs to be limited. The efficacy
and feasibility of using markers with EPIDs was studied

by Balter et al. (32), application of the use of markers
have been extensively studied by Pouliot and co-workers
(33). The use of markers is becoming more popular and
EPID systems are being augmented with software to
detect markers as well as perform the requisite positional calculations.
Quality Assurance and In Vivo Dosimetry
In 2001, Task Group No. 58 of the American Association of
Physicists in Medicine Radiation Therapy Committee
issued a protocol to define the standard-of-care for performing EPID QA on a daily, monthly, and annual basis
(34). In this protocol, a quality assurance program is
proposed where daily checks of EPID system performance, image quality, and safety interlocks are performed
by a radiation therapy technologist. In addition to reviewing results and independently performing checks conducted by the technologist, a medical physicist should
conduct the following checks on a monthly basis: perform
constancy check of SNR, resolution, and localization;
inspect images for artifacts; do a mechanical inspection
of all EPID components; and maintain the computer
system. The annual QA tasks are also performed by the
medical physicist, and include all of the above tasks plus a
full check of the EPID geometric localization accuracy. By
performing these QA tasks, the radiotherapy department
may be reasonably assured of a reliable EPID system for
clinical use.
The QA tests typically utilize a vendor-supplied phantom designed to facilitate evaluation of the aforementioned
tasks. Since clinical linacs are typically dual energy (e.g., 6
and 15 MV) in design, tests are applied to both photon
energies. As expected, the lower photon energy will demonstrate improved image quality (e.g., SNR and spatial resolution). Since many EPID systems utilize sophisticated
computer software utilities, testing of this software in a
realistic setting is an integral component of the EPID
quality management program. As can be expected with
tests that are subjective in nature, it is recommended that
multiple users be employed to evaluate the subjective
criteria so as to minimize user bias.
In addition to the aforementioned advantages, an EPID
system permits unique opportunities of radiotherapy treatment QA. Treatment fields are typically blocked with beam
modifiers to account for irregularities in patient shape.
These beam modifiers include compensating materials
when minor changes are required, or beam wedges when
gross changes are required. Because modern linacs have
features like dynamic wedges and dynamic multileaf collimators, the nonintegral approach that EPIDs offer over
radiographic film (i.e., the ability to obtain several images
at different stages during a dynamic process) permits
continued high-quality QA. Due to the electronic nature
of EPID measurement of the photon energy fluence exiting
a patient, one can perform exit dosimetry and quantitative
comparisons with treatment planning intentions (35).
However, these efforts are currently research driven,
and widespread clinical implementation may not be
expected for a few years.
IMAGING DEVICES
BIBLIOGRAPHY
1. AAPM report No. 24: Radiotherapy portal imaging quality,
1987.
2. Valicenti RK, Michalski JM, Bosch WR, Gerber R, Graham MV,
Cheng A, Purdy JA, Perez CA. Is weekly port filming adequate
for verifying patient postion in modern radiation therapy? Int J
Rad Oncol Biol Phys 1994;30(2):431438.
3. Moseley J, Munro P. Display equalization: A new display
method for portal images. Med Phys 1993;20(1):99102.
4. Van den Heuvel F, Han I, Chungbin S, Strowbridge A, TekyiMensa S, Ragan D. Development and clinical implementation of
an enhanced display algorithm for use in networked electronic
portal imaging. Int J Rad Oncol Biol Phys 1999;45:10411053.
5. Jaffray DA, Batista JJ, Fenster A, Munro P. X-ray scatter in
megavoltage transmission radiography: Physical characteristics
and influence on image quality. Med Phys 1994;21(1):4560.
6. Bailey NA, Horn RA, Kamp TD. Fluoroscopic visualization of
megavoltage therapeutic x-ray beams. Int J Radiat Oncol Biol
Phys 1980;6:935939.
7. Leszczynski KW, Shalev S, Cosby S. A digital video system for
on-line portal verification. Medical Imaging IV: Image Formation. SPIE 1990;1231:401405.
8. Leong J. Use of digital fluorscopy as an on-line verification
device in radiation therapy. Phys Med Biol 1986;31:985992.
9. Munro P, Rawlinson JA, Fenster A. A digital fluoroscopic
imaging device for radiotherapy localization. Int J Rad Oncol
Biol Phys 1990;18:641649.
10. Visser AG, Huizenga H, Althof VGM, Swanenburg BN. Performance of a prototype fluoroscopic radiotherapy imaging
system. Int J Rad Oncol Biol Phys 1990;18:4350.
11. Wong JW, Slessinger ED, Hermes RE, Offutt CJ, Roy T,
Vannier MW. Portal dose images I: Quantitative treatment
plan verification. Int J Rad Oncol Biol Phys 1990;18:1455
1463.
12. Wickman GA. A liquid filled ionisation chamber with high
spatial reslution. Phys Med Biol 1974;19:6672.
13. Meertens H, van Herk M, Weeda J. A liquid ionisation detector
for digitital radiography of therapeutic megavoltage photon
beams. Phys Med Biol 1985;30:313321.
14. Van Herk M, Meertens H. A matrix ionization chamber imaging device for on-line patient set-up verification during radiotherapy. Radiother Oncol 11:369378.
15. Morton EJ, Swindell W, Evans PM. A linear array, scintillation crystal-photodiode detector for megavoltage imaging.
Med Phys 1991;18:681691.
16. Lam KS, Partowmah M, Lam WC. An on-line electronic portal
imaging system for external beam radiotherapy. Br J Radiol
1986;59:10071013.
17. Munro P, Bouius DC. X-ray quantum limited portal imaging
using amorphous silicon flat-panel arrays. Med Phys
1998;25(5):689702.
18. De Neve W, Van den Heuvel F, De Beukeleer M, Coghe M,
Verellen D, Thon L, De Roover P, Roelstraete A, Storme G.
Interactive use of on-line portal imaging in pelvic radiation.
Int J Rad Oncol Biol Phys 1993;25:517524.
19. Van den Heuvel F, De Neve W, Verellen D, Coghe M, Coen V,
Storme G. Clinical implementation of an objective computeraided protocol for intervention in intra-treatment correction
using electronic portal imaging. Radiother Oncol 1995;35:232
239.
20. Balter JM, Pelizarri CA, Chen GTY. Correlation of projection
radiographs in radiation therapy using open curve segments
and points. Med Phys Mar.Apr. 1992;19(2): 329334.
21. Bijhold J, Lebesque JV, Hart AAM, Vijlbrief RE. Maximizing
setup accuracy using portal images as applied to a conformal
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
97
boost technique for prostate cancer. Radiother Oncol

1992;24:261271.
Bel A, Vos PH, Rodrigus PTR, Creutzberg CL, Visser AG,
Stroom JC, Lebesque JV. High-precision prostate cancer irradiation by clinical application of an offine patient setup verification procedure, using portal imaging. Int J Rad Oncol Biol
Phys 1996;35(2).
Langen KM, Jones DT. Organ motion and its management.
Int J Radiat Oncol Biol Phys May 1 2001;50(1):265
278.
Stroom JC, Boer de HC, Huizenga H, Visser AG. Inclusion of
geometrical uncertainties in radiotherapy treatment planning
by means of coverage probability. Int J Radiat Oncol Biol Phys
Mar 1 1999;43(4):905919.
van Herk M, Remeijer P, Lebesque JV. Inclusion of geometric
uncertainties in treatment plan evaluation. Int J Radiat Oncol
Biol Phys Apr 1 2002;52(5):14071422.
ICRU and International Commission on Radiation Units and
Measurements. Prescribing, recording and reporting photon
beam therapy. ICRU Report 1993; 50.
ICRU and International Commission on Radiation Units and
Measurements. Prescribing, recording and reporting photon
beam therapy, Supplement to ICRU 50. ICRU Report 1999;
62.
Erridge SC, Seppenwoolde Y, Muller SH, van Herk M, De
Jaeger K, Belderbos JSA, Boersma LJ, Lebesque JV. Portal
imaging to assess set-up errors, tumor motion and tumor
shrinkage during conformal radiotherapy of non-small cell
lung cancer. Radiother Oncol 2003;66(1):7585.
Jaffray DA, Drake DG, Moreau M, Martinez AA, Wong JW.
Radiographic and tomographic imaging system integrated
into a medical linear accelerator for localization of bone and
soft-tissue targets. Int J Radiat Oncol Biol Phys Oct 1
1999;45(3):773789.
Lattanzi J, McNeeley S, Hanlon A, Schultheiss TE,
Hanks GE. Ultrasound-based stereotactic guidance of
precision conformal external beam radiation therapy in clinically localized prostate cancer. Urology Jan 2000;55(1):
7378.
Serago CF, Chungbin SJ, Buskirk SJ, Ezzell GA, Collie AC,
Vora SA. Initial experience with ultrasound localization for
positioning prostate cancer patients for external beam radiotherapy. Int J Radiat Oncol Biol Phys Aug 1 2002;53(5):1130
1138.
Balter JM, Sandler HM, Lam K, Bree RL, Lichter AS,
ten Haken RK. Measurement of prostate movement
over the course of routine radiotherapy using implanted
markers. Int J Rad Oncol Biol Phys 1995;31(1):113
118.
Vigneault E, Pouliot J, Laverdiere J, Roy J, Dorion M. Electronic portal imaging device detection of radioopaque markers
for the evaluation of prostate position during megavoltage
irradiation: A clinical study. Int J Radiat Oncol Biol Phys
Jan 1 1997;37(1): 205212.
Herman MG, Balter JM, Jaffray DA, McGee KP, Munro
P, Shalev S, van Herk M, Wong JW. Clinical use of
electronic portal imaging: report of radiation therapy
committee task group 58. Med Phys May 2001;28(5):712
737.
Boellaard R, Van Herk M, Mijnheer BJ. A convolution model to
convert transmission dose images to exit dose distributions.
Med Phys 1997;24(2):189199.
See also COMPUTED
TOMOGRAPHY; MAGNETIC RESONANCE IMAGING;
PHOTOGRAPHY, MEDICAL; POSITRON EMISSION TOMOGRAPHY; ULTRASONIC

IMAGING.
98
IMMUNOLOGICALLY SENSITIVE FIELDEFFECT TRANSISTORS
IMMUNOLOGICALLY SENSITIVE FIELDEFFECT

TRANSISTORS
EMMANUEL S. ZACHARIAH
University of New Jersey
New Brunswick, New Jersey
P. GOPALAKRISHNAKONE
National University of Singapore
Singapore
PAVEL NEUZIL
Institute of Bioengineering
and Nanotechnology
Singapore
INTRODUCTION
Diagnostics as a whole represent a large, well-established,
and continually expanding market. Methods for the selective determination of analytes in biological fluids, such as
blood and urine, are important. When a foreign substance
(antigen) invades the human body, the immune system
produces antibodies that interact with the antigen. Such a
recognizing process involves the formation of an immunocomplex based on interactions between the immunospecies. The recognition is specific for the antibody-antigen
system and, thus, for the measurement of antigen concentration (1,2). This determination is of importance for diagnosis because the antigens can be viruses, bacteria that are
involved in many human illnesses, such as cancer and
AIDS. The analytes detected and measured have also
included many other medical diagnostic molecules such
as hormones, clinical disease biomarkers, drugs, and environment pollutants such as pesticides. Antibody diversity is
so great that virtually any biomolecule can be recognized
(3). The range of analyte concentrations encountered is
extremely large, from greater than 103 M for species such
as glucose and cholesterol and to less than 1012 M for
certain drugs and hormones (4). It is for the detection of
these low level analytes that the application of immunological techniques is essential.
An immunoassay is a multistep diagnostic test based on
the recognition and binding of the analyte by the antibody.
Most immunoassay techniques are based on the separation
of free and bound immunospecies (5). In these techniques,
one of the immunoagents (antibody or antigen) is immobilized on a solid phase. The solid phase facilitates the separation and washing steps required to differentiate bound and
free fractions of the label. Quantification of a bound immunoagent is conducted by using labels covalently bound to the
immunoagent with specific properties suitable for detection. The most common labels are radioactive markers,
enzymes, and fluorescent labels. For many of the nonisotopic labels, the reagents have been designed such that
binding of labeled antigen to antibody in some way modulates the activity of the label, resulting in a homogenous
immunoassay without the need for a separation step. The
most familiar type of enzyme immunoassay in clinical
analysis is known as enzyme-linked immunosorbent assay
(ELISA) (6). Different schemes of enzyme immunoassay
exist, and, in clinical laboratory practice, the most popular
are the Sandwich method for large analytes, and compe-
titive binding immunoassay methods for the determination

haptens (low molecular weight analytes).
The advent of biosensor technology, with the possibility
of direct monitoring of immunoreactions, provides opportunity to gain new insight into antigen-antibody reaction
kinetics and create rapid assay devices with wider applications. A biosensor is composed of (1) a biochemical receptor,
which uses biosubstances such as enzymes, antibodies, or
microbes to detect an analyte, (2) a transducer, which
transforms changes in physical or chemical value accompanying the reaction into a measurable response, most
often in the form of electrical signal (79). The term immunosensor is used when antibodies are immobilized to recognize their appropriate antigens (or vice versa) (10).
Immunosensors possess several unique features, such as
compact size, simplicity of use, one-step reagentless analysis, and absence of radioactivity, which make them
attractive alternatives to conventional immunoassay techniques. Immunosensors can be divided, in principle, into
two categories: nonlabeled and labeled (11). Nonlabeled or
direct-acting immunosensors are designed in a way that
the immunocomplex (i.e., the antibody-antigen complex) is
directly determined by measuring physical changes
induced by the formation of the complex. In contrast,
labeled or indirect-sensing immunosensors have incorporated a sensitively detectable label. The immunocomplex is
thus entirely determined through measurement of the
label. In order to determine an antigen, the corresponding
antibody is immobilized on the membrane matrix, which is
held on an amperometric-or potentiometric-sensing transducer used to measure the rate of the enzymatic reaction
(1214).
Of the electrochemical technologies for biosensors, the
Ion-Sensitive Field Effect Transistor (ISFET) has been
the center of special attention as a transducer. ISFETs
were introduced by Bergveld in 1970 (15), and were the
first type of this class of sensor in which a chemically
sensitive layer was integrated with solid-state electronics.
A field effect transistor (FET) can be considered as a
charge-sensitive device (i.e., any change in the excess
interfacial charge at the outer insulator surface will be
mirrored by an equal and opposite charge change in the
inversion layer of the FET). By excluding the gate metal in
a FET and using a pH-sensitive gate insulator, a pHsensitive FET was constructed (16,17). After the invention
of the ISFET, many different types of FET-based sensors
have been presented. The application of enzymes as the
selecting agent in ISFET-based sensing systems leads to
the development of highly sensitive sensors. Such enzymemodified ISFETs (EnFETs) can, in principle, be constructed with any enzyme that produces a change in pH
on conversion of the concerning substrate (18). By combining the ISFET with a membrane that contains a biological
substance, like an antibody, the sensor can detect a specific
antigen (19). The ISFET immunosensors or Immunologically sensitive FETs (IMFETs) have several advantages
over the conventional enzyme immunoassay. The ISFET
could be mass-produced by an integrated circuits (IC)
process, which makes it very small and economical. An
electric circuit can be integrated on the same chip. The
biosensor platform finds many applications in various
99
fields, such as medical diagnostics, fermentation process

control, and environmental monitoring.
THEORY
In order to understand the operation of the IMFET, one
must trace its origins back to the ISFET or ChemFET
(Fig.1). The latter devices have been described in depth
elsewhere (2022). A packaged ISFET is shown in
Fig. 2 (23). ISFETs and ChemFETs have, in turn, evolved
from the Metal Oxide Semiconductor Field Effect Transistor (MOSFET), currently the most popular active device in
the entire semiconductor industry. It is a unipolar device,
where the current is given by the flow of majority carriers,
either holes in PMOS type or electrons in NMOS type. The
operation of the MOSFET can be considered as a resistor
controlled by the status of a gate region, so-called MIS
structure. It is a sandwich consisting of a stacked-gate
Figure 2. Photomicrograph of an ISFET device packaged on PCB
(23).
Figure 1. The hierarchy of field effect transistor. a. MOSFET. b.

ISFET. c. IMFET.
metal layer, an insulator (typically silicon oxide), and a

semiconductor. Assume a low-level doped p-type (NMOS
device). Three different states of charge distribution can
occur, depending on the voltage Vg , applied between the
metal and a semiconductor. A negative value of Vg causes
positive holes to accumulate at the semiconductorinsulator interface. A positive value of Vg of a low magnitude leads to the depletion condition in which mobile
holes are driven away from the interface, resulting in a
negative charge of low density due to the presence of
immobile acceptor atoms. Finally, if the Vg exceeds a
certain threshold voltage (Vth), electrons accumulate at
the semiconductor-insulator interface at a density greater
that the hole density, a situation opposite to that normally
found with p-type semiconductors. This depletion of mobile
charge carriers followed by surface inversion is known as
the field effect. It forms an electrically conductive channel between two other terminals, a source and a drain (see
Figure 1a). The drain current Id through the transistor is a
functions of drain and gate voltage. Without surface inversion (i.e., Vg < Vth,) the drain current is negligible, because
the drain-to-substrate PN junction is reverse biased.
The MOSFET and its descendants are charge-controlled
devices. In analytical applications (e.g., ISFETs, ChemFETs, and IMFETs), the change in charge density is
brought about by adsorption of one or more species present
in the solution onto the FET structure. In the ISFET, the
gate metal is replaced with a conventional reference electrode (Ag/AgCl or Hg/Hg2Cl2), a solution containing an
ionic species of interest, and an electroactive material
(membrane) capable of selective ion exchange with the
analyte (Fig. 1b), which is an example of a nonpolarizable
interface, that is, reversible charge transfer occurs
between the solution and the membrane. The analyte
generates a Nernst potential at the membrane-solution
interface, which then modulates the drain current analogous to the manner in which changing the externally
applied voltage does for the MOSFET.
100
Direct-Acting (Label-Free) IMFET

The structure of the direct-acting IMFET is similar to that
of the ISFET, except that the solution-membrane interface
is polarized rather than unpolarized. If the solution-membrane interface of the ISFET is ideally polarized (i.e.,
charge cannot cross the interface), then the ISFET can
measure the adsorption of charged species at the interface
as shown below. As antibodies, antigens, and proteins are
generally electrically charged molecules, the polarized
ISFET could be used to monitor their nonspecific adsorption at the solution-membrane interface. To render the
polarized ISFET selective for a given antigen and thus
create the so-called IMFET, the specific antibody for that
antigen has to be immobilized on the surface of the ISFET
(see Fig. 1c). The adsorption of this antigen would then be
specifically enhanced over other molecules in the solution
and the signal measured by the ISFET would be mostly due
to the adsorption of that particular antigen. The ISFET
interacts with the analyte through an ion-exchange
mechanism, whereas the IMFET interaction is based on
the antigen-antibody reaction.
This design for the measurement of the adsorption of
charged molecules is practicable only if charge cannot cross
the interface, which, thus, acts as an ideal capacitor. As
will be seen, failure to achieve a perfectly polarizable
interface has a detrimental effect on the specificity of
the IMFET. Few reports exist on direct-acting IMFETs;
a brief analysis on the work of Janata research group will
be presented here (2426). The capacitance of a polarized
interface is described by electrical double-layer theory and
is usually modeled as a series combination of two capacitors, CG and CH, where CG is the capacitance of the diffuse
GouyChapman part of the double layer and CH is the
capacitance of the Helmholtz part of the double layer (27).
The total capacitance, Cdl, is therefore
1=Cdl 1=CG 1=CH
(1)
The electrical circuit through the gate of an ISFET with an

ideally polarized interface can be modeled, therefore, as a
series combination of CG, CH, and C0, as drawn in Fig. 3,
where C0 is the capacitance of the insulator. A gate voltage
VG is applied through a reference electrode between the
solution and the semiconductor. The process of adsorption
of charged molecules can be modeled as the transfer of a
VG
quantity of charge from the solution to the surface of the

transistor as would occur if the switch were closed for a
short time period allowing the current source to transfer
the charge. As adsorption occurs, the charge on each plate
of the capacitors will change to accommodate the new
charge balance. The charge change on capacitor C0 is
the quantity of interest as it represents the charge in
the inversion layer of the FET, Qi, and will affect the drain
current of the transistor, which can be directly measured.
If a quantity of charge, Qads, is transferred by the adsorption of charged molecules, then the charge change on C0,
Qi, can be represented by
Qi Qads fC0 =C0 Cdl g
(2)
Hence, only a fraction of the adsorbed charge will be

mirrored in the transistor. When adsorption occurs,
because electroneutrality must be observed in the system,
an equal quantity of the opposite charge must either enter
the inversion layer of the FET or enter the double layer
from the solution. Equation 2 predicts that part of the
image charge will come from the solution as ions entering
the double layer with the adsorbing molecules. This fraction of charge, which is mirrored in the inversion layer of
the FET, will be defined as b, and it is defined as
b Qi =Qads C0 =C0 Cdl
(3)
According to this model, only 0.3% of the charge on the

adsorbing molecules will be mirrored in the inversion layer
of the FET. The authors conservatively estimated b to be
104. Considering the Id current as a function of the
potential at the solution-membrane interface, it is clear
that a relationship between the adsorbed charge and interfacial potential, FSol-mem, is necessary to describe the chemical response of the IMFET. This potential is merely the
charge change induced in the inversion layer divided by the
insulator capacitance:
FSolmem Qi =C0 bQads =C0
(4)
Substitution of this expression in to Equations 5 and 6

yields the response equations for the polarized ISFET. The
authors derived the following expressions for the polarized
ChemFET relating to Qi to the observed parameter, the
drain current (Id):

mnWC0
V
Vg Vt Er fsolmem d Vd
Id
L
2
Vd < Vdsat
(5)
and
Id
CG
CH
Solution
C0
Semiconductor
Current source
Switch
Figure 3. Electrical model for the measurement of charge

adsorption with the CHEMFET.
mnWC0
Vg Vt Er fsolmem 2
2L
Vd > Vdsat
(6)
where W is the width of the source-drain conducting

channel, mn is the effective electron mobility in the
channel, C0 isthe capacitance per unit area of the gate
insulator, L is the channel length, Vd is the drain-tosource voltage, Vg is the applied gate voltage, Vt is the
threshold voltage (for surface inversion), and Er is the
potential of the reference electrode.
The antibody-antigen binding reaction in its
simplest form can be expressed in terms of the following
biomolecular reaction:
Ab Ag @ AbAg
where Ab is the antibody, Ag is the antigen, and AbAg is the
complex. The reaction is characterized by the equilibrium
constant K,
K AbAg =Ab Ag
(7)
The total charge change at the interface due to the binding,

Qi, can be shown to be
Qi bQads bzF
KAg S
1 Ag
(8)
where z is the ionic charge of the antigen and [S] is the

surface concentration of binding sites (the surface concentration of immobilized antibodies before binding). Substitution of this expression into Equation 4 yields
FSolmem
bzFKAg S
C0 1 Ag
(9)
From Equation 9, the limit and range of the detection for

the IMFET can be predicted. Assume that the equilibrium
constant is in typical range from 105 to 109 (28), which gives
a value of b 104. If the antibodies are immobilized with
a surface concentration of 1 molecule per 10 nm2 and the
charge on an antigen is five electronic charges of an antibody, the IMFETs detection limit would be in the range of
107 1011M of concentration antibody concentration.
The antigen concentration that gives 90% surface coverage
can similarly be calculated to be in the range of
104 108 M. Similar equations can be derived for the
case where the antigen is immobilized at the interface
rather than the antibody. However, it has been argued
by many researchers that a static measurement concerning
the presence of a protein layer on an electrode is difficult,
because the charged groups are, of course, neutralized by
surrounding counter ions (29). In order to avoid interference from other charged species present in the solution, the
substrate for immobilization should preferably be inert
and nonionic (2430), which in aqueous solutions implies
a hydrophobic surface (31). Ideal conditions that are
required in this coherence are a truly capacitive interface
at which the immunological binding sites can be immobilized, a nearly complete antibody coverage, highly charged
antigens, and a low ionic strength.
Schasfoort et al. (32) extensively studied the requirements for the construction of IMFET, which would operate on the direct potentiometric sensing of protein
charges. The charge redistribution around immobilized
proteins at the insulator-solution interface can be
described by the double-layer theory (33). On adsorption,
the diffuse layer of counter ions around the protein
charges may overlap with the diffuse layer of the electrolyte-insulator interface. The thickness of diffuse-charge
layers is described by the Debye theory (34) and defined by
the distance where the electrostatic field has dropped to 1/
e of its initial value:

e ekT 1=2
k1 0 2
2q I
101
where k1 is the Debye length, q the elementary change, k

Boltzmanns constant, T absolute temperature, e0 the
permittivity
of vacuum, e the dielectric constant, and I
P
1=2 ci z2i represents the ionic strength in which ci is the
concentration of ion i with valency z (for 1-1 salt, I can be
replaced by c).
It can be seen from the equation that the Debye length is
strongly dependent on the ionic strength of the solution;
more precisely, the Debye length is inversely proportional
to the square root of the ionic strength. Therefore, one can
expect that the chance of overlapping of the double layers of
the substrate-solution interface and the adsorbed proteins
can be substantial only if low electrolyte concentrations are
used, owing to the dimensions of the proteins (Fig. 4). In a
physiological salt solution, the Debye length is limited to
ca. 0.8 nm. It is obvious that only charge density changes
that occur within the order of a Debye length of the ISFET
surface can be detected. With the macromolecules, such as
protein, the dimensions are much larger (about 10 nm)
than those of the double layer of the electrolyte-insulator
interface, which means that, in such a case, most of the
protein charge will be at a distance greater than the Debye
length from the surface. If, moreover, on top of a monolayer
of antibody molecules a second layer on antigens in
coupled, it is obvious that the chance of overlap of the
diffuse layers of antigens with electrolyte-substrate interface will decrease even more. At high ionic strength, the
additional charges of the antigen are nearly always located
far outside the diffuse layer at the ISFET surface and pure
electrostatic detection of these antigenic charges, therefore, is impossible. In addition, a theoretical approach is
Figure 4. Schematic representation of the potential distribution

in a direct-acting IMFET. K1 is the Debye length; dAb, dimension
of macromolecule (e.g., antibody).
102
given based on the Donnan equilibrium description, which

provides an insight into the potential and ion distribution
in the protein layer on the IMFET (32). It is shown that the
Donnan potential and the internal pH shift, induced by the
protein charges, compensate each other to a greater extent.
If the ISFET shows Nernstian behavior, it can be concluded
that a direct detection of protein charge is impossible. In
order to construct an IMFET, a reference FET or ISFET
with a low sensitivity would satisfy the detection of the
partially compensated Donnan potential in the presence of
an adsorbed protein layer. However, the application of such
as IMFET is limited to samples with low ionic strength.
An alternative, indirect approach is proposed by Schasfoort et al. (35,36) for the detection of an immunological
reaction taking place in a membrane, which covers the gate
area of an ISFET (Figs. 5a and 5b). The protein layer on the
gate is exposed to pulse-wise increases in electrolyte concentration. As a result, ions will diffuse into the protein layer
and, because of a different mobility of anions and cations,
transients in potential will occur at the protein-membrane
solution interface. The ISFET, being a voltage-sensitive
device, is suitable for the measurement of these transients.
As the mobility of ions is a function of the charge density in
the protein membrane, changes in the charge density will
influence the size and direction of the transients. By exposing the ISFET to a pH gradient and a continuous series of ion
concentration pulses, the isolectric point of the protein layer
can be detected and, thus, changes as the result of an
immunological reaction. When a membrane separates two
(mV)
C2
C1
2
Time (s)
Figure 5. An ion-step arrangement: an ISFET exposed to an

increased electrolyte concentration. Transient potential can
bemeasured, developing from transient transport of ions across
the membrane, which is caused by stepwise changes in
electrolyte concentration. The ISFET response f as a result of
the stepwise changes in electrolyte concentration (C1C2).
compartments with different electrolyte concentrations, a

potential gradient can be measured. The different diffusion
rates of anions and cations through the membrane set up a
static membrane potential, which can be expressed by the
NernstPlanck equation (33):
fm
RT
a
D D
:U:In 2 U
F
a1
D D
where fm the membrane potential, RT and F have their

common meaning, U the ratio of the diffusion coefficients
(D and D) of cations and anions, and a1 and a2 are the
electrolyte activities in the respective compartments. The
ion-step method is further developed by Schasfoort and
Eijsma (37), and a detailed theoretical understanding of
the ion-step response has been presented by Eijiki et al. (38).
Recently, an impedance spectroscopy method was used
tocharacterize immobilized protein layers on the gate of
an ISFET and to detect an antigen-antibody recognition
event (39).
Indirect-Sensing IMFET
Although the ion-step method is an indirect way of measuring antigen-antibody reaction that occurs on the gate
region of an ISFET, it does not involve any reagents that
enhance or amplify the signal intensity. Many approaches
to transduction of the antibody-antigen combining event
are indirect. They are based on the ability of an enzyme
label to produce electroactive substances within a short
span of time. Antibody or antigen is immobilized on the
gate area of pH-FET. In the competitive binding assay, the
sample antigen competes with enzyme-labeled antigen for
the antibody-binding sites on the membrane. The membrane is then washed, and the probe is placed in a solution
containing the substrate for the enzyme. IMFETs based on
the sandwich assay are applicable for measuring large
antigens that are capable of binding two different antibodies. Such sensors use an antibody that binds analyteantigen, which then binds an enzyme-labeled second antibody. After removal of the nonspecifically adsorbed label,
the probe is placed into the substrate-containing solution,
and the extent of the enzymatic reaction is monitored
electrochemically. Gate voltage is supplied by reference
electrode, such as Ag/AgCl or a Hg/Hg2Cl2 electrode, that is
immersed in a sample solution. It is, however, difficult to
make a small conventional electrode, which prevented the
IMFET from being miniaturized as a whole. When a noble
metal, such as platinum or gold, is used as a reference
electrode, the potential between the metal electrode and
sample solution fluctuates. The fluctuation makes stable
measurement impossible. A method to cancel the fluctuation using a reference ISFET (REFET) is reported. A combination of two kinds of ISFET is used, one of which detects a
specific substance whereas the other (REFET) does not
detect it (Fig. 6). Thus, measuring the differential output
between the two ISFETs can cancel the potential fluctuation
in the sample solution and drift due ISFET (4042).
Most of the indirect-sensing IMFET studies are carried
out using urease-conjugated antibodies. Urea is used as a
substrate. The immunosensor uses a reaction wherein urea
is hydrolyzed by the urease-labeled second antibody. The
103
PRACTICE
Direct-Acting IMFET
Figure 6. Differential measurement setup for an IMFET.
reaction is

H2 NCONH2 2H2 O H ! 2NH
4 HCO3
According to the reaction, the pH value in the membrane

becomes high. On the other hand, on the ISFET surface
with inactive antibody membrane, the above reaction does
not occur and pH remains constant. Hence, by measuring
the differential output between two ISFETs, only pH
changes due to urea hydrolysis are detected. In some cases,
the authors used antibodies conjugated with the glucose
oxidase. These sensors use oxidation of glucose by glucose
oxidase. In the reaction, gluconic acid is produced and the
pH value in the glucose oxidase immobilized membrane
becomes low. To achieve a high sensitivity of horseradish
peroxidase (HRP) detection, various substrates, either
alone or in combination, are tested and the result is shown
in Fig. 7.
pH-change ( pH)
0.8
0.6
2
0.4
0.2
H2O2
0
3
4
0
20
40
60
80
100
Time, S
Figure 7. Typical ISFET responses for HRP (109 M) substrates
(1) Ascorbic Acid O-phenylenediamine (OPD); (2) OPD; (3)
piodophenol luminol; (4) catechol.
The rationale for attempting to combine the fields of

immunology and electrochemistry in the design of analytical devices is that such a system should be sensitive due to
the characteristics of the electrochemical detector while
exhibiting the specificity inherent in the antigen-antibody
reaction. The ideal situation would be to detect the binding
of immunoreagents directly at an electrode, for example, by
changes in surface potential, which could be truly
described as an immunosensor (43,44). Much more effort
has been committed to develop transducers, which rely on
direct detection of antigen by the antibody immobilized on
its surfaces (or vice versa). In 1975, Janata immobilized a
sugar-binding protein Concanavalin A on a PVC-coated
platinum electrode and studied its responses in the presence of sugar (30). The potential of the electrode with
respect to an Ag/AgCl electrode changed owing to adsorption of the charged macromolecule. Although the system
reported was not based on an immunochemical reaction,
the finding of a potentiometric response stimulated further
investigations in this field. Direct potentiometric sensing of
antigen human choriogonadotropin (hCG) with an antihCG antibody sensitized titanium wire resulted in 5 mV
shifts with respect to a saturated calomel electrode (45).
The change in potential was explained by a simple charge
transfer model.
In 1978, Schenck first proposed a concept of direct
immunosensing by an ISFET (46,47). He suggested using
FET with, on the gate region, a layer of antibody specific to
a particular antigen. Replacement of electrolyte solution
with another electrolyte solution-containing antigen
should alter the charge of the protein surface layer due
to the antigen-antibody reaction, thus affecting the charge
concentration in the inversion layer of the transistor. The
corresponding change in the drain current would then
provide a measure of the antigenic protein concentration
in the replacement solution. Many research groups have
tried to realize the proposed concept of Schenck, but the
results obtained are meager (48,49). Collins and Janata
immobilized a PVC membrane containing cardiolipin antigen onto the gate of a previously encapsulated ChemFET
(50). They demonstrated that the solution-membrane
interface was somewhere between a polarized and a nonpolarized interface, based on the measured membrane
exchange current density. The measured potential was
therefore a mixed potential deriving out of the permeation
of Na and Cl ions into and out of the membrane. The
change in potential following specific binding of antibody to
the membrane was due primarily to a perturbation of the
mixed potential, rather than to the adsorbed charge from
the antibody itself. Therefore, the device could not be
considered selective for the immunoreactive species of
interest. Besides, Janata reported that it is impossible to
construct an IMFET without having an ideal polarized
solution-insulator interface. He proclaimed all of his earlier results as artifacts (51). In spite of these practical
difficulties, Gotoh et al. (52) published results obtained
with an IMFET sensitive to Human serum albumin (HSA).
104
Figure 8. Outline of an Ion-step flow through system.
A 2 mV shift was detected with HSA containing polyvinylbutyral membrane deposited on an ISFET after reaction
with its antibody. It appears that experimental results
obtained with direct detection of protein on solid-state
electrode or similar devices are, so far, limited to secondorder effects. Nevertheless, a real theoretical explanation
is absent. Therefore, until more experimental evidence is
available, the true value of direct-acting IMFET concept
remains to be established.
Schasfoort et al. (36) proposed an alternative approach to
overcome the above-described difficulties of a direct detection of immunological reaction with ISFET. By stepwise
changing the electrolyte concentration of the sample solution, a transient diffusion of ions through the membraneprotein layer occurs, resulting in a transient membrane
potential, which can be measured by the ISFET. A flowthrough system was used to carry out the experiments as
schematically drawn in Fig. 8. The pH of the electrolyte can
be changed by using a gradient vessel. When the solution
flows under hydrodynamic pressure out of vessel 1, the pH
will change through mixing with a solution of different pH
from vessel 2. By opening the value for a few seconds, the
ISFET can be exposed to a higher salt concentration. The
step change in join concentration was completed within
50 ms. After 2 s the valve was closed and the membrane
can gain equilibrate with the buffer flowing out of vessel 1.
In order to exchange the electrolyte concentration rapidly,
the volume between the valve and the ISFET was kept
small. ISFETs with a polystyrene-agarose membrane were
incubated with 105 M HSA for 3 h. The ISFET response
was measured as a function of the pH, and the inversion
point was determined to be pI 3.72 0.05. Subsequently,
the ISFETs were incubated in different concentrations of

anti-HSA antibodies solution ranging from 0.06 to 64 mM.
The anti-HSA antibody was able to change the inversion
point of the HSA-coated membrane from 3.70 to 5.55. The
above experiments clearly demonstrated that the net
charge density in a protein layer deposited on an ISFET
could be determined by exposing the membrane to a stepwise change in electrolyte concentration while measuring
ISFET current change. The transient membrane potential
observed is a result of the different mobilities of the positive
and negative ions present in the protein layer. It is also
observed that characteristic inversion points and slope are a
function of the protein concentration and type of protein.
Also isolectric points could be detected from the membrane
potentials as a function of the pH. This detection of the
isoelectric point of a protein complex is the basis for the
development of an IMFET. An immunological reaction
results in a change of the fixed-charge density in the membrane, which can be explained by a shift of the protein
isoelectric point due to the immunological reaction.
The ion-step method was originally designed to measure
immunoreaction via the change in charge density, which
occurs in an antibody-loaded membrane, deposited on an
ISFET, upon reaction with a charged antigen. The efficacy
of ion-step method for the quantification of a non-charged
antigen was demonstrated using progesterone as the model
analyte (53). Progesterone is an uncharged molecule,
hence, it cannot be detected directly by using the ion-step
method. A competitive method was devised using a charged
progesterone-lysozyme conjugate. To prepare the ISFETs
for ion-step measurement, a membrane support was
created by depositing a 1:1 mixture of polystyrene beads
and agarose on the gate. The ISFETs were then cooled to
4 8C and the solvent was slowly evaporated, leaving a
porous membrane with a thickness of approximately 4
mm. The ISFET was then heated to 55 8C for 1 h to
immobilize the membrane onto the gate. The ISFET was
placed in the flow-through system (see Fig. 8) and a monoclonal antibody specific to progesterone was incubated on
the membrane (0.5 mg/ml, 4 8C for 20 h). A competitive
assay method was used to detect progesterone levels, and
the detection limit was approximately 108 M of progesterone in the sample solution. Recently, Besselink et al.
(54) described an amino bead-covered ISFET technology
for the immobilization of antibodies. HSA was immobilized
onto the amino bead-coated ISFET, by covalent cross-linking method, and the anti-HSA antibodies were quantitated
using the ion-step method. The antibody concentration was
detected within 15 min, with yields up to 17 mV (Fig. 9).
Indirect-Sensing IMFET
The indirect-sensing IMFET concept emerged during the
early 1990s in order to overcome the difficulties met with
the direct-acting IMFET devices (55). Colapicchioni et al.
(56) immobilized IgG using protein A onto the gate area of
an ISFET. The efficacy of the IMFET was demonstrated
using Human IgG and atrazine antibodies captured using
protein A. As the atrazine is a small molecule (hapten),
which does not induce an immunoresponse as such, it was
linked to a carrier protein. Bovine Serum Albumin (BSA)
Potential (mV)
10
15
20
0.00
0.10
0.20
0.30
0.40
0.50
Time (s)
Figure 9. Ion-step responses of HSA-coated ISFET before (upper
solid curve) and after incubation (for 15 min) with undiluted antiHSA (lower solid curve) and anti-BSA (dashed curve). Ion stepping
was performed at pH 4.02.
was conjugated to ametryn sulfoxide, which has structural

similarity with atrazine, and the ametryn-BSA conjugate
was injected into rabbit to raise antibodies. A sandwich
assay format was used to detect Human IgG and a competitive assay format was used to quantitate atrazine concentration. The antigen-antibody reaction was monitored
by the addition antihuman IgG-GOD conjugate and
ametryn-GOD, respectively. Glucose was used as the substrate and the pH variation was detected by the ISFET.
The sensitivity of the assay was 0.1 mg/ml and 1 ppb for
human IgG and atrazine, respectively. An ISFET-based
immunosensor was demonstrated for the detection of bacterial (Clostridium thermocellum) cells. The analysis
included the reaction of antibacterial antibodies with cells
in suspension or after covalent immobilization of cells on
porous photoactivated membranes and, subsequently, the
revelation of bound antibodies by the conjugate of protein A
and HRP and the quantitation of enzyme activity with
ISFET. The sensitivity of the sensor was within a range
of 104107 cells per ml (57). Selvanaygam et al. (23)
reported ISFET-based immunosensors for the qunatitation
of b-Bungarotoxin (b-BuTx), a potent presynaptic neurotoxin from the venom of Bungarus multicinctus. A murine
monoclonal antibody (mAb 15) specific to b-BuTx was
immobilized on the gate area, and the antigen-antibody
reaction was monitored by the addition of urease-conjugated rabbit anti-b-BuTx antibodies. The sensor detected
toxin level as low as 15.6 ng/ml. The efficacy of the sensor
for the determination of b-BuTx from B. multicinctus
venom was demonstrated in the mouse model.
An immunological Helicobacter pylori urease analyzer
(HPUA), based on solid-phase tip coated with a monoclonal
antibody specific to H. pyloris urease and ISFET, was
reported by Sekiguchi et al. (58). A solid-phase tip,
with an inner diameter of 0.55 mm, coated with the monoclonal antibody, was incubated for 15 min at room temperature in an endoscopically collected gastric mucus
sample. The activity of urease captured on the inner surface of the solid-phase tip was measured by coupling it with
an ISFET in a measuring cell containing urea solution. The
pH change of urea solution after 55 s of the enzymatic
reaction inside the tip was measured by withdrawing 1.1 ml

of solution toward the upstream of the tip, where the
measuring ISFET was installed. One cycle of measurement
was completed in 17.5 s, and the sensitivity of system was
0.2 m IU/ml. The calibration curve for the quantitation of
urease is shown in Fig. 10. Clinical studies were carried out
with 119 patients (75 males and 44 females with an average age of 51, ranging from 13 to 79) who underwent
gatroduodenoscopy and judged necessary to evaluate the
infection of H. pylori and urea breath test (UBT) was used
as a gold standard. Thirty-three of the UBT positive 36
patients were positive, and 81 of UBT negative 83 patients
were negative by HPUA resulting in the 92% sensitivity
and 98% specificity.
An IMFET for the detection of HIV-specific antibodies
based on a combination of ELISA principle and ISFET flow
injection analysis setup was presented by Aberl et al. (59).
The active sensing components consist of a reaction cartridge containing a carrier with the immobilized receptor
layer and an ISFET sensor mounted in a flow-through cell.
A flow cell was constructed using two ISFET sensors on
one in a two-channel configuration (Fig. 11). The liquid
2.4
2.2
2.0
1.8
1.6
1.4
1.2
pH
105
1.0
0.8
0.6
0.4
0.2
Cut off: pH =0.010
0.01
0.1
10
100
1000 10000
Final concentration of Hp urease (mIU/ml)

Figure 10. A standard curve for HPUA.
106
Sample
Valve
Reaction cartridge
Carrier
solution
Reference
solution
ISFET 1
ISFET 2
Roller pump
Waste
Reference
electrode
Waste
Figure 11. Diagrammatic representation of a flow injection
system for indirect immunosensing.
headspace on top of the ISFET sensors was reduced to

about 1 ml, and the dead volume of the whole sensor cell
was 7ml. The detection principle was realized according to
the sandwich ELISA procedure using urease as a pH
shifting marker enzyme. Antigen molecules (p24 or
gp120) were immobilized on cellulose nitrate membranes
mounted in a special flow-by cartridge or the inner surface
of Borosilicate glass capillary tubing. After blocking the
unspecific binding sites, the antigen was reacted with
specific serum in different dilution or nonspecific serum
as a negative control. In comparison with conventional
ELISA, the ISFET-FIA ELISA showed a slight lower sensitivity. The antibodies were detected in a serum diluted
more than 1:12,000 in ELISA, whereas the sensitivity of
the ISFET FIA ELISA was between a 1:1000 and a
1:10,000 dilution. Glass as a support material showed
highly reproducible test results when compared with cellulose nitrate membrane.
Tsuruta et al. (60) reported a fully automated ISFETbased ELISA system using a pipette tip as a solid phase and
urease as a detecting enzyme. The inner wall of the end
part of a pipette tip was used as a solid phase, and the
urease activity of the conjugate, captured after a two-step
immunoreaction, was measured by coupling the pipette
tip with the ISFET in a pH measuring cell (Fig. 12). A twostep sandwich assay procedure was used for the quantitation of AFP, CEA, HBsAg, and HBsAb, and a two-step
competition assay was used for HBcAb, and secondantibody configuration was used for HTLV-1 Ab. After
final incubation in conjugate solution, the pipette tip
was washed and it was introduced into the pH measuring
cell in order to couple it with ISFET. At the same time,
feeding of the substrate solution was stopped, to read the
pH change for 20 s. The output (source potential) of the
ISFET was read and stored in the CPU during the abovementioned 20s at 0.1 s intervals. The maximum changing
rate of the source potential (DV/Dt, mV/s) was calculated
from these 200 data points. The total assay time was 21 min
as the sum of 5, 10, 5 and 1 min for preheating of sample,
First immunoreaction, Second immunoreaction, and pH
measurements, respectively. The assay speed was 60 samples/h. Assay performance, such as within run CVs,
between run CVs, detection limits, and correlation with
the conventional ELISA kits, were satisfactory for all of six
Figure 12. Cross-sectional view of a pH-measuring cell.
analytes. The detection limit for CEA, 0.09 mg/l was comparable to better than those reported for the most advanced
chemiluminescent ELISA system (0.086 mg/l).
Polymerase chain reaction (PCR) has proven to be of
great importance in clinical diagnosis (61). Usually, the
PCR products have been detected by staining with ethidium
bromide in qualitative methods, and fluorescent dyes in
real-time quantitation. Although electrophoresis has the
advantage of giving information on the molecular size of
PCR products, it is not well-suited to mass screening or
automation. On the other hand, real-time monitoring is
well-suited for mass screening and automation but is expensive. One of the most promising methods for automatizing
the detection system of PCR products is ELISA. An ISFETbased ELISA was used to quantitate PCR products (62).
Double-stranded PCR products with digoxigenin and biotin
at both terminals were obtained by using digoxigenin-and
biotin-labeled primers. The PCR products were detected by
a two-step sandwich ELISA. One ml of the solution after PCR
was introduced into the end part of the solid-phase pipette
tip coated with antidigoxigenin antibody. Biotin-labeled
PCR products captured at the solid phase were detected
with avidin-urease conjugate, and the enzyme activity was
measured by the ISFET in a pH measuring cell containing
urea solution. The detection limit of the system was determined using a known amount of purified PCR product
labeled with digoxigenin and biotin, and it was found that
10 amol of the labeled DNA in 1 ml sample. The assay was
used to detect HTLV-1 provirus gene integrated in the
genome of human MT-Cell, and it was found that 100 pg
of the genomic DNA was specifically detectable after 35
cycles of PCR. The apparent dynamic range for detection of
MT-1 DNA was from 100 pg to 100 ng.
One of the most important targets in molecular biology
is the quantitation of mRNA related to special disease by
RT-PCR. The accuracy of quantitative RT-PCR has been
remarkably improved by the introduction of competitive
RT-PCR, in which a synthetic RNA is used as an internal

standard (63). Tsuruta et al. (64) developed a ISFETELISA method for the qantitatiion of mRNA in clinical
samples. In this method, a fixed amount of a synthetic
RNA, pRSET RNA, was added as internal standard to the
solution of target RNA (IL-1b) after reverse transcription,
PCR was carried out using digoxigenin-labeled sense primer and biotin-labeled antisense primer for IL-1b, and
FITC-labeled sense primer and a biotin-labeled antisense
primer for pRSET. The double-stranded PCR products of
IL-1b and pRSET were captured by two solid-phase pipette
tips, one coated with antidigoxigenin antibody and another
with anti-FITC antibody, respectively, and sandwiched by
an avidin-urease conjugate, whose activity was measured
with ISFET. The ratio of the signal intensity for IL-1b to
that for pRSET was used to quantitate the concentration of
IL-1b. A calibration curve was obtained using a known
amount of AW109 RNA as an external standard in place of
IL-1b m RNA. It was found that 102106 copies of IL-1b
mRNA were measurable by the present method. Expression levels of IL-1b mRNA in clinical samples, such as
monocytes of peripheral blood or synovial cells from
patients with RA or OA, were determined.
Practical Limitations
In this section, we shall address some practical problems
that have been limiting factors in the commercial application of IMFETs. The widespread use of IMFETs for applications ranging from medical diagnosis to industrial
process control or environmental monitoring has not actually happened. The underlying reasons for this situation
fall into two main categories, those that are inherent to the
transistor, such as material, encapsulation, and reference
electrode, and those problems common to its application as
an immunosensor function, such as, antibody immobilization, stability, and durability. The pH sensing properties
and drift behavior of the ISFET is the main limiting factor
in the commercial breakthrough of ISFET. After the invention of the ISFET, initially the only gate material used was
SiO2. Although SiO2 showed pH, sensitivity of 20 to 40 mV/
pH, the thermally grown gate oxide loses its isolation
property within a few hours of immersion in a solution.
In order to isolate this gate oxide from the solution, another
isolating layer, such as Si3N4, Al2O3, or Ta2O5, has to be
placed on top of this gate oxide. A layer of Si3N4 on top of
SiO2 showed 4550 mV/pH, and other layers, such as Al203
and Ta2O5, showed even higher pH sensitivity, 5357 mV/
pH and 5559 mV/pH, respectively (65). Drift rate for Si3N4
is reported as 1 mV/h and for Al2O3 and Ta2O5 0.10.2 mV/
h after 1000 min of operation at pH 7.0. In most of the work
on IMFETs published so far, these three gate materials,
Si3N4, Al2O3, and Ta2O5, have been used. IMFETs are also
sensitive to light and temperature (66).
The pH-sensitive ISFETs can be fabricated by means of
standard MOS technology, except for the metallization
step. However, after dicing the wafers into single chips,
the substrate becomes exposed at the edges of the senor.
Encapsulation and packaging are two final processing
steps that determine reliability and durability (lifetime)
of the IMFETs. In order to achieve high quality sensors, all
107
electrical components have to be isolated from their surroundings. Several reports exist on the encapsulation and
packaging of ISFET devices for pH application (21). The
simplest method of isolating these sides is encapsulation
with epoxy-type resins. The most important ISFET characteristics, such as stability, accuracy, and durability, also
pertain to the reference electrode. One of the major hurdles
in IMFETs is the lack of a solid-state reference electrode.
The small IMFETs have to be combined with a conventional KCl-solution-filled reference electrode. In order to
achieve miniaturized IMFET, it is important to miniaturize the reference electrode. In general, two approaches
have been followed: reference FETs (REFETs), which are
used in an ISFET/REFET/quasi-reference electrode setup,
and miniaturized conventional reference electrodes. In the
first approach, attempts have been made to cover the
ISFET surface with a pH-insensitive layer or to render
the surface pH insensitive by chemical modification. In the
second approach, the structure of a conventional electrode
(mostly Ag/AgCl type) is miniaturized partially or completely on a silicon wafer. Its potential is a function of concentration of chloride ions. They are supplied either from
an internal electrolyte reservoir formed by an anisotropic
etching in the silicon wafer or by adding chloride ions into
the test solution.
Some of the technological factors such as pH sensitivity
and drift can now be overcome with the existing technology. A hurdle peculiar to direct-acting IMFET is the need to
provide a thin but fully insulating layer (membrane)
between the antigen or antibody coating and the semiconductor surface. Such a membrane must be thin enough to
allow a small charge redistribution occurring as a result of
analyte (antigen-antibody) binding to exert a detectable
change in electrical field. Conversely, it must also provide
adequate insulation to prevent dissipation of the field by
leakage of ions. Even assuming that the ideal insulating
membrane can be developed, a further hurdle may need to
be overcome. Surface charges and hydrogen binding sites of
proteins cause a counter-ion shell (double-layer) and structured water shells to surround the molecules; these regions
of structured charge will inevitably contribute to the electrical field affecting the FET gate. Pending these breakthroughs, the development of direct-acting IMFETs
appears to be stagnant.
The immobilization methods used for immunosensors
include a variety of adsorption, entrapment, cross-linking,
and covalent methods. In general, a covalent immobilization method consisting of silanization step and subsequent
coupling procedure via glutaraldehyde has been used to
immobilize antibodies onto the gate region (67,68). However, no methodical investigation about antibody stability,
storage, and lifetime exists. Reproducible regeneration of
the sensing area is one of the major problems met with
IMFETs that have been used for continual monitoring or
repeat usage. The need for renewal of the sensing surface
derives from the high affinity constants derived from the
strong antigen-antibody reaction. Two different strategies
have been used to achieve the renewal of the sensing surface, breakage of the antigen-antibody bond and reusing the
immunologic reagent immobilized on the solid phase. A
second alternative is the elimination of antigen-antibody
108
complex from the solid support and immobilization of fresh

immunologic material. Dissociation of antigen from antibody is usually carried out in low pH and high ionic strength
solutions. Protein A was chemically immobilized onto the
gate surface by using a polysiloxane layer of [3-(2-aminoethyl)aminopropyl]trimethoxysilane
(APTES)
and
cross-linking agent such as glutaraldehyde. Reversibility
of the linkage between Protein A and antibodies in order to
restore the device for the next measurement was studied by
breaking the antibody-antigen complex formed on Protein A
using a variety of reagents. Glycine buffer pH 2 and 3 and
MgCl2 3.5 M were found to be more effective when compared
with other tested reagents due to high ionic strength (55).
Selvanayagam et al. (23) studied the reusability of an
ISFET sensor by removing the antibody membrane from
the gate area. The regenerated devices tested were reported
to function normally five times, although a considerable
amount of time was required for the regeneration process.
Recently, IMFET using magnetic particle and integrated to
flow injection system has been described to overcome the
problem of regeneration (69,70). The immunological material was immobilized on the surface of magnetic particles
and were transported by a flow system, and were retained
on the gate area of the ISFET by a magnetic field produced
by a magnet (Fig. 13). The regeneration of immunologic
materials was achieved by releasing the magnetic field,
thus freeing those particles that were washed by the flow
system, and new magnetic particles were injected and
retained on the surface of transducer by reacting the magnetic field. A fresh and reproducible surface was thus produced, ready for the next analytical cycle.
The main barrier to the successful introduction of
IMFETs for clinical testing is undoubtedly the high performance and automation level of the machines that
already exist in centralized laboratories. They have been
developed specifically for use with either immunoassay or
clinical chemistry. Immunoassay performance is continually being optimized and assay times have been reduced
over the past few years. Depending on the parameter, the
assay time can be as low as 6 min and the majority of the
larger machines could carry out between 100 to 200 testes
per hour. IMFETs must be compared with these methods
with respect to assay time, sensitivity, and cost. The need
for in-built calibration has been frequently encountered in
sophisticated quantitative IMFETs. Although feasible and
acceptable in laboratory-based instrumentation, it remains
H2 N-CO-NH 2
H 2O
Ureaseconjugate
2 nd antibody
CO 2 + 2 NH 3
Rabbit IgG
Magnetic particle
Magnet
ISFET
Figure 13. Schematic representation of the magnetoIMFET.
a major problem in small disposable IMFET devices. To

facilitate the increased use of IMFETs, one should look for
real tests and standards that prototypes can meet, based on
current diagnostic needs and perceived future development. The progress of IMFET beyond the experimental
laboratory level is mainly dependent on how skillfully its
development and marketing are combined with parameter
selection.
FUTURE DIRECTIONS
A key consideration in antibody immobilization to the gate
area is to maintain, reproducibly, the highest possible
binding activity after immobilization while conserving
the amount of antibody used. However, many aspects, both
fundamental and more applied, require in-depth study
before IMFETs can become successful commercial device,
including improved control of biomolecule immobilization
and novel immobilization strategies; enhancement of biomolecule stability and retention of activity in vitro; and the
ability to reproduce the high signal-to-noise ratios obtained
in simple test solutions in real samples such as blood or
water. The IMFETs tends to respond nonspecifically to any
molecule bound to the surface; hence, it affects the measurement parameter to some extent. The specificity of
analyte detection, therefore, relies entirely on achieving
high ratios of specific to nonspecific binding, which, in the
context of low concentrations of analyte in blood, can
represent a formidable problem. Reduction of nonspecific
binding is another area that will continue to be of major
importance. The ability to immobilize ordered antibodies
will maximize antigen binding to a given surface while
reducing the availability of nonbinding site sections of the
immobilized antibody, or uncovered surface areas, which
can promote nonspecific interaction with other components
in the sample. The potential advantages of using Fab
fragments rather than more complex intact antibodies
(such as IgG) could be explored.
IMFETs, similar to immunoassays, involve multistep
procedures, including washing steps, incubation periods,
and quite complex signal generation protocols. It is likely
that research efforts into a novel signal amplification
system, without the normally associated complications
of multi-reagents or multistep protocol will be of increasing importance. The irreversibility of antigen-antibody
interaction presents a major challenge in designing
IMFETs for continual monitoring or repeated usage.
Treatment of an antibody-antigen complex with a mildly
denaturing medium for a short time interval has shown
some promise in regenerating sensor surfaces. Development of enhanced denaturation conditions, which optimize dissociation of antigen while minimizing irreversible
loss of antibody structural integrity, may be possible in
the near future. The use of catalytic antibodies in immunosensors has been proposed. The ability of catalytic
antibodies to catalyze the hydrolysis of phenyl acetate
with the formation of acetic acid allows integration of pHsensitive microelectrodes to give a potentiometric immunosensing system (71). The advantage of catalytic antibodies over normal antibodies is that reversibility of
response is readily achieved, because bound antigen

reacts to form a product with a low affinity for the antibody, resulting in dissociation. As the binding event is
followed immediately by a catalytic reaction and release of
the reaction products, the molecular recognition site is
regenerated with each molecular reaction; as a consequence, catalytic antibodies can be used to create reversible IMFETs for continuous monitoring of analyte
concentrations. Improvements in sensitivity and crossreactivity are likely to be achieved as a result of the
increasing interest in this field of research.
CONCLUSION
Although the ISFET concept has existed for over 30 years,
its practical applications such as the IMFETs are still
emerging very slowly. The relatively slow rate of progress
of IMFET technology from inception to fully functional
commercial devices for these applications is a reflection of
technology-related and market factors. In general, two
approaches have been followed in the past to realize
IMFETs. In the first approach, antigen-antibody reaction
on an immonbilized membrane was monitored without
any addition of labels. The second approach takes the
advantage of an enzyme label to indirectly monitoring
the antigen-antibody reaction using pH-sensitive FET.
The development of IMFETs that directly detect
antigen-antibody reaction is extremely challenging;
only a few examples exist, the majority of which are
without valid theoretical explanation. Although it shows
enormous promise in the early stages of development, an
effective, reliable, and analyte-selective direct-acting
IMFET sensor is yet to be constructed. The ion-step
method represents a novel measurement concept for
potentiometric detection and quantification of an
adsorbed antigen or antibody molecule in which modified
ISFETs are used. Many approaches to transduction of the
antibody-antigen combining event are indirect, necessarily involving the use of reagents admixed with analyte,
and therefore cannot be seen as routes to development of
True IMFETs. Nevertheless, such reagent-dependent,
indirect-sensing IMFETs may offer real commercial
advantages over the current generation direct-acting
IMFET readout technologies. The clinical diagnostic field
offers real opportunities for the exploitation of IMFET,
but because it is a highly competitive and well-established
market, those who wish to introduce new products must
carefully target their market niche. IMFETs will have to
compete with such technology on the basis of factors such
as cost, ease of use, sensitivity, operational stability,
robustness, and shelf-life.
BIBLIOGRAPHY
1. Catty D. editor, Antibodies Volume 1. A Practical Approach.
Washington, DC: IRL; 1988.
2. Mayforth RD. Designing Antibodies. New York: Academic
Press; 1993.
3. Johnstone AP, Turner MW. Immunochemistry 1. A Practical
Approach. New York: IRL; 1997.
109
4. Bluestein BI, Walczak IM, Chen SY. Fiber optic evanescent

wave immunosensors for medical diagnostics. Trends Biotechnol 1990;8:161168.
5. Gosling JP, A decade of development in immunoassay methodology. Clin Chem 1990;36:14081427.
6. Kemeny DM, A Practical Guide to ELISA. New York: Pergamon Press; 1991.
7. Turner APF, Karube I, Wilson GS. editors. Biosensors: Fundamentals and Applications. New York: Oxford University
Press; 1987.
8. Buerk DG, Biosensors: Theory and Applications. Lancester,
PA: Technomic; 1993.
9. Eggins BR. Biosensors: An Introduction. New York: Wiley;
1996.
10. Aizawa M. Principles and applications of electrochemical and
optical biosensors. Anal Chim Acta 1991;250:249256.
11. North JR. Immunosensors: Antibody-based biosensors.
Trends Biotechnol 1985;3:180186.
12. Aizawa M. Immunosensors for clinical analysis. Adv Clin
Chem 1994;31:247275.
13. Morgan LC, Newman DJ, Price CP. Immunosensors: Technology and opportunities in laboratory medicine. Clin Chem
1996;42:193209.
14. Ghindilis AL, Atanasov P, Wilkins M, Wilkins E. Immunosensors: Electrochemical sensing and other engineering
approaches. Biosens Bioelecton 1998;13:113131.
15. Bergveld P. Future applications of ISFETs. Sens Actuators B
1991;4:125133.
16. Kuriyama T, Kimura J. An ISFET biosensor. In: Wise DL,
editor, Applied Biosensors. Boston, MA: Butterworth; 1989.
17. Kimura J, Kuriyama T. FET biosensors. J Biotechnol
1990;15:239254.
18. Van der Schoot BH, Bergveld P. ISFET based enzyme sensors.
Biosensors 1987/88;3:161186.
19. Yuquing M, Jianguo G, Jianrong C. Ion-sensitive field effect
transducer-based biosensors. Biotechnol Advances 2003;
21:527534.
20. Janata J, Moss SD. Chemically sensitive field-effect transistors. Biomed Eng 1976;11:241245.
21. Bergveld P, Sibbald A. Comprehensive Analytical Chemistry,
XXIII: Analytical and Biomedical Applications of Ion-Selective
Field-Effect Transistors. New York: Elsevier; 1988.
22. Bergveld P. Thirty years of ISFETOLOGY. What happened in
the past 30 years and what may happen in the next 30 years.
Sens Actuators B 2003;88:120.
23. Selvanayagam ZE, Neuzil P, Gopalakrishnakone P, Sridhar
U, Singh M, Ho LC. An ISFET-based immunosensor for the
detection of b-Bungaraotoxin. Biosens Bioelectron 2003;
10:405414.
24. Janata J, Huber RJ. Chemically sensitive field effect transistors. In: Freiser H, editor. Ion-Sensitive Electrodes in Analytical Chemistry, vol. II, New York: Plenum Press; 1980.
25. Janata J, Blackburn GF. Immunochemical potentiometric
sensors. Ann N Y Acad Sci 1984;428:286292.
26. Blackburn GF. Chemically sensitive Field effect transistors.
In: Turner APF, Karube I, Wilson S. editors. Biosensors:
Fundamentals and Applications. Oxford: Oxford Science Publications; 1987.
27. Bockris JOM, Reddy AKN. Modern Electrochemistry, vol. 2.
New York: Plenum Press; 1970.
28. Eisen HN. Immunology. An Introduction to Molecular and
Cellular Principles of the Immune Response. New York:
Harper and Row; 1974.
29. Bergveld P. A critical evaluation of direct electrical protein
detection methods. Biosens Bioelecton 1991;6:5572.
30. Janata J. An immunoelectrode. J Am Chem Soc 1975;97:2914
2916.
110
31. Janata J, Janata J. Novel protein-immobilizing hydrophobic

polymeric membrane, process for producing same and apparatus employing same. U.S. Patent 3,966,580, 1976.
32. Schasfoort RBM, Bergveld P, Kooyman RPH, Greve J. Possibilities and limitations of direct detection of protein changes by
means of immunological field-effect transistor. Anal Chim
Acta 1990;238:323329.
33. Moore WJ. Physical Chemistry. London: Longman; 1976.
34. Davies JT, Rideal EK. Interfacial Phenomena. London: Academic Press; 1963.
35. Schasfoort RBM, Bergveld P, Bomer J, Kooyman RPH, Greve
J. Modulation of the ISFET response by an immunological
reaction. Sens Actuators 1989;17:531535.
36. Schasfoort RBM, Kooyman RPH, Bergveld P, Greve J. A new
approach to ImmunoFET operation. Biosens Bioelecton 1990;
5:103124.
37. Schasfoort RBM, Eijsma B. The ions-step method applied to
disposable membranes for the development of ion-responding
immunosensors. Sens Actuators 1992;6:308311.
38. Eijkel JCT, Olthuis W, Bergveld P. An ISFET-based dipstick
device for protein detection using the ion-step method. Biosens
Bioelectron 1997;12:9911001.
39. Kharitonov AB, Wasserman J, Katz E, Willner I. The use of
impedance spectroscopy for the characterization of proteinmodified ISFET devices: Application of the method for the
analysis of biorecognition processes. J Phys Chem 2001; B105:42054213.
40. Sergeyeva TA, Soldatkin AP, Rachkov AE, Tereschenko MI,
Piletsky SA, Elskaya AV. b-Lactamase label-based potentiometric biosensor for a-2 interferon detection. Anal Chim Acta
1999;390:7381.
41. Yacoub-George E, Wolf H, Koch S, Woias P. A miniaturized
ISFET-ELISA system with a pre-treated fused silica capillary
as reaction cartridge. Sens Actuators B 1996;34:429434.
42. Starodub NF, Dzantiev BB, Starodub VN, Zherdev AV. Immunosensor for the determination of the herbicide simazine based
on an ion-selective field-effect transistor. Anal Chim Acta
2000;424:3743.
43. Wang J. Electroanalytical Techniques in Clinical Chemistry
and Laboratory Medicine, New York: VCH; 1988.
44. Luppa PB, Sokoll LJ, Chan DW. Immunosensors principles
and applications to clinical chemistry. Clin Chim Acta
2001;314:126.
45. Yamamoto N, Nagasawa Y, Sawai M, Sudo T, Tsubomura H.
Potentiometric investigations of antigen-antibody and
enzyme-enzyme inhibitor reactions using chemically modified
metal electrodes. J Immunol Methods 1978;22:309317.
46. Schenck JF. Technical difficulties remaining to the application
of ISFET devices. In: Cheung PW, editor. Theory, Design and
Biomedical Applications of Solid State Chemical Sensors. New
York: CRC; 1978.
47. Schenck JF. Field effect transistor for detection of biological
reactions. U.S. Patent 4,238, 757, 1980.
48. Schoning MJ, Poghossiana A. Recent advances in biologically
sensitive field-effect transistors (BioFETs). Analyst 2002;
127:11371151.
49. Janata J, Thirty years of CHEMFETs a personal view.
Electroanalysis 2004;16:18311835.
50. Collins S, Janata J. A critical evaluation of the mechanism of
potential response of antigen polymer membrane to the corresponding antiserum. Anal Chim Acta 1982;136:9399.
51. Janata J. Proceedings of the 2nd International Meeting on
Chemical Sensors; 1986:2531.
52. Gotoh M, Tamiya E, Karube I. Micro-FET biosensors using
polyvinylbutyral membrane. J Membrane Sci 1989;41:291
303.
53. Schasfoort RBM, Keldermans CEJM, Kooyman RPH, Bergveld P, Greve J, Competitive immunological detection of progesterone by means of the ion-step induced response of an
immunoFET. Sens Actuators 1990;BI:368372.
54. Besselink GAJ, Schasfoort RBM, Bergveld P. Modification of
ISFETs with a monolayer of latex beads for specific detection
of proteins. Biosens Bioelecton 2003;18:11091114.
55. Toshihide K. Electrochemical sensors in immunological measurement. E.U. Patent 0,328,380, 1998.
56. Colapicchioni C, Barbaro A, Porcelli F, Giannini I. Immunoenzymatic assay using CHEMFET devices. Sens Actuators
B 1991;6:186191.
57. Akimeno VK, Khomutov SM, Obraztsova AY, Vishnivetskii
SA, Chuvilskaya NA, Laurinavichus KS, Reshetilov AN. A
rapid method for detection of Clostridium thermocellum by
field-effect transistor immunodetection. J Microbiol Methods
1996;24:203209.
58. Sekiguchi T, Nakamura M, Kato M, Nishikawa K, Hokari K,
Sugiyama T, Asaka M. Immunological Helicobacter pylori
urease analyzer based on ion-sensitive field effect transistor.
Sens Actuators 2000;B67:265269.
59. Aberl F, Modrow S, Wolf H, Koch S, Woias P. An ISFET-based
FIA system for immunosensing. Sens Actuators 1992;B6:186
191.
60. Tsuruta H, Yamada H, Motoyashiki Y, Oka K, Okada C,
Nakamura M. An automated ELISA system using a pipette
tip as a solid phase and a pH-sensitive field effect transistor as
a detector. J Immunol Methods 1995;183:221229.
61. Saiki R, Scharf S, Faloona F, Mullis K, Horn G, Erlich H,
Arnheim N. Enzymatic amplification beta-globulin genomic
sequences and restriction analysis for diagnosis of sickle cell
anemia. Science 1985;230:13501354.
62. Tsuruta H, Matsui S, Hatanaka T, Namba T, Miyamoto K,
Nakamura M. Detection of the products of polymerase chain
reaction by an ELISA system based on an ion-sensitive field
effect transistor. J Immunol Methods 1994;176:4552.
63. Wang AM, Doyle MV, Mark DF. Quantitation of mRNA by the
polymerase chain reaction. Proc Natl Acd Sci USA 1989;
86:97179721.
64. Tsuruta H, Matsui S, Oka K, Namba T, Shinngu M, Nakamura M. Quantitation of IL-1-beta mRNA by a method of RTPCR and an ELISA based on ion-sensitive field effect transistor. J Immunol Methods 1995;180:259264.
65. Gopel W, Hesse J, Zemel JN editorss. Sensors: A Comprehensive Survey, Chemical and Biochemical Sensors, Part I. Verlagsgesellschaft Mbh: VCH; 1991.
66. Neuzil P. ISFET integrated sensor technology. Sens Actuators
1995;B24-25:232235.
67. Filippo P, Antonio NC. Sensor with antigen chemically bonded
to a semiconductor device. E.U. Patent 0,395,137, 1990.
68. Starodub NF, Samodumova IM, Starodub VN. Usage of organosilanes for integration of enzymes and immunocomponents
with electrochemical and optical transducers. Sens Actuators
1995;B24-25:173176.
69. Santandreu M, Sole S, Fabregas E, Alegret S. Development of
electrochemical immunosensing systems with renewable surfaces. Biosens Bioelectron 1998;13:717.
70. Sole S, Alegret S, Cespedes F, Fabregas E. Flow injection
immunoanalysis based on a magnetoimmunosensor system.
Anal Chem 1998;70:14621467.
71. Blackburn GF, Talley DB, Booth PM, Durfor CN, Martin MT,
Napper AD, Rees AR. Potentiometric biosensor employing
catalytic antibodies as the molecular recognition element.
Anal Chem 1990;62:22112216.
See also ION-SENSITIVE
FIELD EFFECT TRANSISTORS.
IMMUNOTHERAPY
IMMUNOTHERAPY
QIAO LI
University of Michigan
Ann Arbor, Michigan
INTRODUCTION
Immunotherapy of cancer, infectious disease, and autoimmune disease has opened a new area for disease management. This approach has developed very fast lately due to
the advances and involvements of modern technology in
molecular biology, cell biology, immunology, biochemistry,
and bioengineering. Adoptive T cell immunotherapy of
cancer involves passive administration of lymphoid cells
from one host to another, or back to itself in order to
transfer tumor immunity for cancer treatment. It was first
realized >20 years ago that adoptive immunotherapy
may be feasible to treat human malignancies. However,
the early form of this practice was quite simple. It could be
as easy as a straightforward blood cell transfer. The apparent inefficiency of antitumor immune responses, and the
failure to successfully combat the disease laid the foundation for current concepts of immunotherapy. It did not take
too long before it was realized that boosting the antitumor
immune response by deliberate vaccination could increase
the potential benefits of immune cell-based therapies. In
addition, activation of lymphoid cells with monoclonal
antibodies (mAb) toward the molecules involved in T cell
signaling pathways has resulted in therapeutic effector T
cells. The use of immune adjuvants coadministrated with
the cell infusion has enhanced the antitumor efficacy of the
transferred cells and has made adoptive cellular immunotherapy a promising strategy for cancer treatment. Studies on the trafficking of adoptively transferred cells in vivo
as well as the identification and characterization of T cell
subsets responsible for antitumor reactivity have provided
valuable insights toward the development of novel immunotherapeutic strategies. The adoptive immunotherapy of
established tumors with the transfer of tumor-reactive
lymphoid cells has now been shown to be highly effective
against significant tumor burdens both in animal models
and in clinical trials. This is, at least in part, due to recent
developments in this area, such as treating cancer in
special settings (e.g., in lymphopenic hosts induced by prior
conditioning); redirecting the effector cells to tumor
through genetic engineered chimerical T cell receptors
(TCRs) or by transferred tumor antigen-specific TCRs;
and the use of these strategies in combination. This article
intends to review the above developments that have made
adoptive T cell immunotherapy an attractive alternative
for cancer treatment.
INDUCTION OF TUMOR-REACTIVE PRE-EFFECTOR
T CELLS IN VIVO
Successful induction of tumor-reactive pre-effector cells
in a tumor-bearing host represents the first step toward the
conduct of an effective adoptive T cell immunotherapy of
cancer. This procedure provides a source of pre-effector
cells for subsequent T cell activation and expansion in vitro
111
to generate large numbers of effector cells to be infused

back to the tumor-bearing host or cancer patient for
therapy. Due to the relative lack of immunogenicity and
potential immunosuppressive mechanisms of human
malignancies, application of tumor T cell therapy in the
clinical setting has been hampered for a long time by
difficulties to reliably isolate tumor-sensitized lymphoid
cells from the cancer-bearing host. Nevertheless, recent
observations in animal studies and clinic trials have led to
the development of strategies to induce T cell sensitization
in vivo.
Peripheral blood lymphocytes (PBL) represents a convenient source of pre-effector cells. However, in most cases,
particularly in the case of solid tumors, the frequency of
tumor-specific pre-effector cells in PBL is extremely low,
generally far below what is observed in response to viral
infection. Experimental studies and clinical experience
with adoptive immunotherapy have demonstrated that
tumor-draining lymph node (TDLN) cells are potentially
effective antitumor reagents. Chang et al. was the first to
evaluate vaccine-primed LN (VPLN) as a source of lymphoid cells that could be secondarily sensitized by in vitro
methods to generate effector cells capable of mediating
regression of established tumors upon adoptive transfer
in clinical trials (13). These trials included subjects with
metastatic melanoma, renal cell cancer, and head and neck
squamous cell cancers, and have resulted in prolonged,
durable, complete responses.
In murine models, it has been observed that TDLN
harbor lymphoid cells that are functionally capable of
mediating rejection of immunogenic tumors in adoptive
transfer after in vitro activation (4,5). However, both
tumor-infiltrating lymphocytes (TIL) and TDLN cells were
found to be incapable of mediating the regression of poorly
immunogenic tumors such as the B16BL6 melanoma, a
highly invasive tumor of spontaneous origin. It was then
discovered that the subcutaneous inoculation of B16BL6
tumor cells admixed with the bacterial adjuvant, Corynebacterium parvum, resulted in reactive TDLN cells that
differentiated into therapeutic effector T cells upon activation in vitro (6,7). Upon adoptive transfer, these LN cells
successfully mediated the regression of established
tumors. In addition to the ability to mediate regression
of experimentally induced pulmonary metastases, these
activated cells were effective in the treatment of spontaneous visceral metastases originating from a primary
tumor, a condition that more closely approximates human
malignancy. These studies thus demonstrated that vaccination of animals with irradiated tumor cells admixed with
a bacterial adjuvant was capable of inducing tumorreactive T cells in the draining LN.
We have applied these methods to generate vaccineprimed LN in patients with advanced melanoma and renal
cell cancer (RCC) for therapy (3,8). Patients with RCC or
melanoma received intradermal inoculation of irradiated
autologous tumor cells admixed with Bacillus Calmette
Guerin (BCG) as a vaccine. Seven to ten days later, draining LN were removed for in vitro activation and expansion.
Activated LN cells were then administrated intravenously
with the concomitant administration of IL-2 for immunotherapy with defined success (3).
112
IMMUNOTHERAPY
Studies demonstrated that tumor cells genetically modified with immunostimulatory genes are capable of sensitizing T cells. Transfection of cytokine genes into murine
tumor cells have resulted in reduced tumorigenicity following inoculation of the modified tumor cells into animals (9).
In these studies, animals that rejected the inoculum of
modified tumor cells also rejected a subsequent challenge
of unmodified parental tumor cells, thus demonstrating the
development of tumor immunity. We performed a clinical
study of patients with melanoma to evaluate the immunobiological effects of GMCSF transduced autologous tumor
cells given as a vaccine to prime draining lymph nodes (10).
There was an increased infiltration of dendritic cells (DCs)
in the GMCSF-secreting vaccine sites compared with the
wild type (WT) vaccine sites. This resulted in a greater
number of cells harvested from the GMCSFVPLNs compared with the WTVPLNs. Patients received adoptively
transferred GMCSFVPLN cells secondarily activated
and expanded in vitro. A complete clinical response was
observed in one of five patients. This work documented
measurable immunobiologic differences of GMCSFtransduced tumor cells given as a vaccine compared with
WT tumor cells.
Collectively, these observations suggested that TDLN
or VPLN cells may represent an ideal source of tumorreactive T cells. They also established the rationale for
developing tumor vaccines utilizing autologous tumors
admixed with bacterial adjuvant or genetically modified
with cytokine genes, which may prove useful in facilitating
the generation of immune T cells for adoptive immunotherapy.
ACTIVATION AND POLARIZATION OF EFFECTOR

T CELLS IN VITRO
A major challenge in T cell immunotherapy of cancer is how
to activate and expand the relatively low numbers of
tumor-specific T cells obtained from the tumor-bearing
host. Previous studies demonstrated that freshly isolated
TDLN cells had defects in TCR-mediated signal transduction and were not immediately competent in adoptive
transfer models (11,12). It has therefore become a critical
prerequisite in adoptive immunotherapy to expand the preeffector cells into large numbers of effector cells while
augmenting their antitumor reactivity.
In vitro T cell activation using monoclonal antibodies in
the absence of antigen takes advantage of common signal
transduction pathways that are ubiquitous to T cells. This
principle has been used to expand tumor-primed T cells
contained within TDLN or VPLN. The initial efforts
involved the use of anti-CD3 mAb as a surrogate antigen
to activate tumor-primed lymphoid cells, followed by
expansion in IL-2 (12). This approach resulted primarily
in the generation of CD8 effector cells that mediated
tumor regression in vivo. Subsequent clinical studies utilizing this method to activate VPLN cells demonstrated
that this cellular therapy can result in achieving durable
tumor responses in subjects with advanced cancer (1,3). We
have extended these investigations in animal models and
with human samples by examining other mAbs that deliver
costimulatory signals in concert with anti-CD3 to activate

tumor-primed lymphoid cells. These other antibodies have
involved anti-CD28 and anti-CD137 (1316). The results of
these investigations have indicated that costimulation can
increase the proliferation of tumor-primed lymphoid cells
and their ability to mediate tumor regression in vivo.
Several important principles in animal models that are
relevant for the treatment of human malignancies have
been identified. For example, the in vitro cytokine profiles
released by effector T cells when cocultured with tumor
cells are found to be predictive of their ability to mediate
tumor regression in vivo. Effector cells that mediate a type
1 (i.e., IFNg) and GMCSF response to tumor antigen are
capable of eradicating tumor upon adoptive transfer. In
contrast, cells that demonstrate a type 2 profile (i.e., IL-10,
IL-4) appear suppressive, and do not mediate tumor regression (16,17). We have determined the importance of IFNg
in mediating tumor regression both in animal studies (16)
and in clinical trials (3). In a phase II adoptive cellular trial
in patients with advanced renal cell cancer, we demonstrated that IFNg secretion and the IFNg: IL-10 ratio of
cytokine released by effector T cells in response to tumor
antigen was associated with clinical outcomes. Specifically,
activated T cells that have an increased IFNg:IL-10 ratio
correlated with tumor response (3). Although effector T
cells can be generated through antibody activation to
mediate tumor regression in animal models, clinical
responses in adoptive immunotherapy have been confined
to a minority of patients. One potential reason for these
limited responses is that antibody-activation procedures
generally stimulate T cells broadly without discriminating
between type1 and type 2 cells, presumably due to the
polyclonal expansion characteristics of antibodies directed
to the TCR common chain, for example, CD3e of the TCR/
CD3 complex or CD28. As a result, both type 1 cytokines,
such as IL-2, IFNg, and type 2 cytokines, for example,
IL-4, IL-5, and IL-10, are modulated (13,18). Therefore,
alternative protocols need to be defined that will preferentially stimulate the type 1 cytokine profile to generate
more potent tumor-reactive T cells for cancer immunotherapy. Toward this end, various in vitro strategies have been
investigated utilizing additional signaling stimuli to promote Th1/Tc1 cell proliferation and antitumor reactivity
(19,20). We reported that costimulation of TDLN cells
through newly induced 4-1BB and CD3/CD28 signaling
can significantly increase antitumor reactivity by shifting
T cell responses toward a type 1 cytokine pattern, while
concomitantly decreasing type 2 response (16). Using the
proinflammatory cytokines, we recently reported that IL12 and IL-18 can be used to generate potent CD4 and
CD8 antitumor effector cells by synergistically polarizing
antibody-activated TDLN cells toward a Th1 and Tc1
phenotype, and that the polarization effect was NF-kB
dependent (21).
The recognition and use of cell polarization strategies
during and /or post antibody activation of T cells represents
another significant change and addition to the traditional
practice of adoptive therapy. While adoptive immunotherapy of cancer requires large numbers of therapeutic T cells
for transfer into cancer patients, the phenotype and cytokine profile of these cells are crucial in determining the
IMMUNOTHERAPY
outcomes of the therapy. The use of polarizing reagents to

modulate T cell function toward the type 1 response provides a rational strategy to enhance the efficacy of cellular
therapy.
USE OF IMMUNE ADJUVANT IN CONCERT

WITH T CELL ADMINISTRATION
In the course of adoptive immunotherapy of cancer, administration of T cell growth factors accompanying T cell transfer may promote T cell activation, proliferation, and tumor
killing, and therefore augment clinical outcomes for the
therapy. These growth factors, as well as other immune
modulatory reagents used in concert with T cell transfer,
function as immune adjuvants in eliciting antitumor activities in vivo. The most useful adjuvant to T cell transfer
to date has been the exogenous administration of IL-2
(13,22,23).
Nearly 20 years ago, Rosenberg and colleagues performed a pilot protocol to investigate the feasibility and
practicality of immunotherapy of patients with advanced
cancer using TIL and recombinant IL-2 (22). The study
represents an initial attempt to use TIL plus IL-2 administration with enhanced tumoricidal capacity in the adoptive immunotherapy of human malignancies. Twelve
patients with melanoma, renal cell carcinoma, breast carcinoma, or colon carcinoma were treated with varying
doses and combinations of TIL, IL-2, and cyclophosphamide. Three partial responses (PR) to therapy were
observed. No toxic effects were directly attributable to
TIL infusions. However, the toxicities of therapy were
similar to those ascribed to IL-2. Indeed, the use of IL-2
has resulted in significant morbidity associated with cellular therapies (3,24). Moreover, a few recent studies
showed that IL-2 may negatively regulate effector cells
through activation-induced cell death (23,25), expanding
the frequency of CD4CD25 T cells, or cause cell redistribution secondary to Ag-induced cell death (26,27). These
studies suggest that novel reagents need to be identified to
serve as alternative immune adjuvants for adoptive T cell
therapy.
In a recent study (25), failed adoptive T cell therapy
could be reversed with low dose IL-15 administration, but
not IL-2. A related T cell growth factor, IL-15, protected T
cells against activation-induced cell death and promoted
homeostatic maintenance of memory T cells and, therefore,
may be advantageous to T cell-based cancer treatment.
Similarly, the role of IL-15 in early activation of memory
CD8 CTLs has been described (28). In this study, memory
CD8 T cells expressing OVA-specific TCR were transferred into IL-15-transgenic (Tg) mice, IL-15 knockout
(KO) mice, or control C57BL/6 mice followed by challenge
with recombinant Listeria monocytogenes expressing OVA
(rLM-OVA). In vivo CTL activities were significantly
higher in the IL-15 Tg mice, but lower in the IL-15 KO
mice than those in control mice at the early stage after
challenge with rLM-OVA. In vivo administration of rIL-15
conferred robust protection against reinfection via activation of the memory CD8 T cells. In addition, IL-27 is a
novel IL-12 family member that plays a role in the early
113
regulation of Th1 initiation and synergizes with IL-12 in

IFNg production (29). Mechanistic studies revealed that
although a comparable proliferative response to IL-27 was
observed between STAT1-deficient and wild-type CD4 T
cells, synergistic IFNg production by IL-27 and IL-12 was
impaired in STAT1-deficient CD4 T cells. IL-27 also
augmented the expression of MHC class I on CD4 T cells
in a STAT1-dependent manner (29).
Although the in vivo administration of proinflammatory
cytokines has demonstrated antitumor efficacy, their
potent antitumor activity is often achieved at the expense
of unacceptable toxicity. For example, IL-12 and IL-18
administration was found to be associated with lethal
organ damages, attributed in part to extremely high levels
of host-induced IFNg production (30). It is anticipated that
administration of low doses of proinflammatory cytokines
in the context of passively transferred TDLN cells will lead
to increased therapeutic efficacy. To this end, the adjuvant
effect of low dose cytokine administration over a long
period of time can be compared with that of a high dose
over a short period of time. These experiments should help
to determine if prolonged administration of low dose cytokines can enhance the therapeutic efficacy by improving
trafficking, survival, and proliferation of the adoptively
transferred T cells.
While toxicity of traditionally used IL-2 limits its clinical utility at high doses, use of novel cytokines at tolerable
low doses in conjunction with cellular therapy may provide
alternative strategies that are less toxic. If the newly
identified proinflammatory cytokines, such as IL-15 and
IL-27 prove to be useful adjuvants to T cell therapy, they
may result in more effective antitumor responses with
reduced morbidity.
TRAFFICKING AND PROLIFERATION OF EFFECTOR

T CELLS AFTER ADOPTIVE TRANSFER
Adoptive T cell therapy has been used for treatment of viral
and malignant diseases with encouraging results. However, little is known about the fate and trafficking of the
transferred effector cells. A study performed at NCI
assessed the trafficking of gp100-specific pmel-1 cells to
large, vascularized tumors that express or do not express
the target Ag (31). It was found that approximately equal
numbers of pmel-1 T cells infiltrated the Ag-positive and
-negative tumors. Massive infiltration and proliferation of
activated antitumor pmel-1 cells were observed in a variety
of peripheral tissues, including lymph nodes, liver, spleen,
and lungs, but not peripheral blood. However, T cell function, as measured by production of IFNg, release of perforin, and activation of caspase-3 in target cells, was
confined to Ag-expressing tumor. It was thus concluded
that adoptively transferred CD8 T cells traffic indiscriminately and ubiquitously while mediating specific tumor
destruction.
We recently characterized the infiltration of adoptively
transferred TDLN cells in the host bearing pulmonary
metastases (21). The TDLN cells were activated with
anti-CD3/anti-CD28 followed by cell culture in IL-12
IL-18 before transfer into tumor-bearing host. The TDLN
114
IMMUNOTHERAPY
cells were labeled with CFSE immediately before infusion.

Immunohistochemical evaluation of adoptively transferred
TDLN cells accumulating in pulmonary tumor nodules was
performed. Infused TDLN cells were observed to (1) attach
to venules, (2) mix with host leukocytes in perivenular
collections, and (3) infiltrate tumor nodules. Active migration of infused cells into pulmonary tumor nodules was
found to be correlated with significant tumor regression.
This corroborates a previous report by Plautz et al. showing
that infused TDLN cells must infiltrate pulmonary nodules
to suppress tumor growth (32).
Several other reports support the hypothesis that efficient tumor regression needs the in situ accumulation of
transferred effector cells. Another study conducted at the
University of Michigan demonstrated that the infused cells
must accumulate in metastatic lesions to suppress tumor
growth, and that the process is dynamic (33). In studies
treating murine lung metastases with adoptively transferred TDLN cells, the TDLN donor cells were initially
confined to alveolar capillaries with no movement into
metastases after infusion. However, within 4 h, TDLN
cells began migrating across pulmonary postcapillary
venules and first appeared within metastases. After 24 h,
most donor cells in the lung were associated with tumor
nodules. Donor cell proliferation both within the lung and
in the lymphoid organs was detected. Importantly, T cells
that had proliferated in the lymphoid organs trafficked
back to the tumor-bearing lungs, accounting for 50% of
the donor cells recovered from these sites. These studies
demonstrate that adoptively transferred TDLN cells
migrate directly into tumor-bearing organs and seed the
recirculating pool of lymphocytes after infusion. Cells that
have differentiated in lymphoid organs eventually migrate
into the tumor site. Additionally, in vitro-generated MelanA-specific CTLs were found to survive intact in vivo for
several weeks and localize preferentially to tumor (34).
Over all, these studies suggest that methods to improve
trafficking and recruitment of donor T cells to the tumor
may improve therapeutic efficacy of cellular therapy.
The availability of congenic strains of mice bearing T
cell markers that differ by epitopes that can be identified by
monoclonal antibodies allows us to track adoptively transferred cells in a semisyngeneic host. In order to perform
quantitative tracking studies of the infused cells, the congenic strain of B6 mouse that expresses CD45.1 can be used
to generate TDLN for transfer into CD45.2 hosts. Analysis
of the infiltrate can be performed by mechanical dissociation of the tumors in order to recover viable lymphoid
infiltrates. By FACS analysis, the number of transferred
CD4/CD8 T cells can be quantified. Proliferation of infused
cells can be assessed by labeling them with CFSE. Confirmed correlation between effective tumor regression and
the infiltration of infused cells to tumor should encourage
further attempts to modulate T cell trafficking by biochemical controls or by genetic modification of well-identified
adhesion molecules, for example, LFA, ICAM, and selectins. Furthermore, a very recent study described the regulation of T cell trafficking by sphingosine 1-phosphate
(S1P) receptor 1 (S1P1) (35). Mature T cells from S1P1
transgenic mice exhibited enhanced chemotactic response
toward S1P, and preferentially distributed to the blood
rather than secondary lymphoid organs, such as draining

lymph nodes. This work suggests that S1P1 affects
systemic trafficking of peripheral T cells, and therefore
makes the S1P/S1P1 signaling pathway a novel target for
T cell trafficking modulation.
IDENTIFICATION AND CHARACTERIZATION OF T CELL

SUBSETS RESPONSIBLE FOR ANTITUMOR REACTIVITY
The CD8 CTLs have long been recognized as the effector
cells that mediate tumor regression. In addition, CD4
effector T cells and NK cells have also been identified to
directly or indirectly mediate tumor regression. We
reported that CD28 costimulation of tumor-primed lymphoid cells promotes the generation of potent tumor-reactive effector cells, particularly CD4 T cells. These antiCD3/anti-CD28 activated CD4 TDLN cells could independently mediate tumor regression in adoptive immunotherapy (13,14,21).
It has to be presumed that any source of antitumor
reactive T cells derived from the tumor-bearing host, that
is, TDLN, will represent a small percentage of the total
population of retrieved cells. Therefore, a theoretically
practical approach would be the identification, isolation,
activation and expansion of subsets of T cells capable of
mediating tumor regression. In this endeavor, Shu and coworkers found that the down-regulation of the homing
molecule L-selectin could serve as a surrogate marker for
the isolation of specific tumor-sensitized T cells (18). In
adoptive immunotherapy of established intracranial MCA
205 tumors, L-selectinlow (CD62Llow) cells displayed at least
30-fold greater therapeutic efficacy than unfractionated
cells. The L-selectinhigh cells did not demonstrate any antitumor effects. These results demonstrate that the purification of L-selectinlow cells led to the generation of immune
effector cells with unusually high therapeutic efficacy
against chemically induced tumors. After that, Plautz
et al. used advanced tumor models in a stringent comparison of efficacy for the L-selectinlow subset versus the total
population of TDLN cells following culture in high dose IL2. L-selectinlow subset comprised 57% of the TDLN cells.
Adoptive transfer of activated L-selectinlow cells eliminated
14-day pulmonary metastases and cured 10-day subcutaneous tumors, whereas transfer of maximally tolerated
numbers of unfractionated TDLN cells was not therapeutic
(36). At the same time, it was identified that tumor-induced
high
L-selectin
cells were suppressor T cells that mediated
potent effector T cell blockade and caused failure of otherwise curative adoptive immunotherapy (37). The treatment failure using unfractionated TDLN cells was due
to cotransfer of the L-selectinhigh suppressor T cells present
in TDLN. However, the L-selectinhigh suppressor T cells
were only found in day-12 TDLN. In contrast, day-9 TDLN
and normal spleens lacked L-selectinhigh cells.
It was not long before a second surrogate marker was
identified for the isolation of tumor-specific T cells. Stoolman et al. described that tumor-specific responses in TDLN
were concentrated in cells expressing P-selectin ligand
(Plighigh T cells) (38). This study found that the minor
subset of TDLN T cells expressing binding sites for the
IMMUNOTHERAPY
adhesion receptor P-selectin (Plighigh T cells) produced

T lymphoblasts with the most tumor-specific IFNg synthesis in vitro and antitumor activity following adoptive
transfer in vivo. The cultured Plighigh TDLN cells were
10- to 20-fold more active against established pulmonary
micrometastases than cultured, unfractionated TDLN, and
>30-fold more active than cultured TDLN cells depleted of
the Plighigh fraction. The Plighigh T cells expressed high
levels of CD69 and low levels of CD62L (L-selectinlow),
which agrees with the previous studies on L-selectin in
TDLN. Further supporting these observations is a recent
study indicating that recruitment of IFNg-producing cells
into the inflamed retina in vivo is preferentially regulated
by P-selectin glycoprotein ligand (39).
In a different attempt to selectively activate tumorsensitized T cells, superantigens were utilized in vitro to
stimulate effector cell generation in a murine model (40).
The TDLN cells stimulated with staphylococcal enterotoxins A (SEA), B (SEB) or C2 (SEC2) resulted in the selective
expansion of Vb3 and 11, Vb3 and 8, or Vb8.2 T cells,
respectively. Adoptive transfer studies revealed that SEBand SEC2-, but not SEA- stimulated cells mediated tumorspecific regression. These results suggested that T cells
bearing Vb8 may preferentially respond to the growing
tumor than T cells bearing Vb3 or 11 elements of the T cell
receptor. Similarly, stimulating TDLN cells with different
anti-Vb mAbs instead of the pan-T cell reagent anti-CD3
mAb enabled the selective activation of Vb T cell subsets
(17). Enrichment of Vb subsets of TDLN cells revealed that
Vb8 cells released high amounts of IFNg and GMCSF
with minimal amount of IL-10 in response to tumor, and
mediated tumor regression in vivo. In contrast, enriched
population of Vb5, Vb7, and Vb11 cells released low
amounts of IFNg and GMCSF with high levels of IL-10,
and had no in vivo antitumor reactivity. In vitro depletion
of specific Vb subsets from the whole TDLN pool confirmed
that the profile of cytokine released correlated with in vivo
antitumor function. These studies indicate that functional
Vb subpopulations of effector cells express differential
antitumor reactivity, and that selective stimulation of
tumor-sensitized T cells is feasible and may represent a
more efficient method of generating therapeutic T cells for
therapy.
Application of cell subsets for successful T cell therapy
should include two approaches: identification of T cell
subsets responsible for mediating antitumor reactivity as
discussed above, and simultaneously, the elimination of
those subsets that are non-reactive or even suppressive.
Characterization of regulatory CD4CD25 T cell subpopulation in terms of their potential suppressive effects on
anticancer effector cells would warrant further investigations in this area. A current study showed that CD8 T cell
immunity against a tumor self-antigen is augmented by
CD4 T helper cells, but hindered by naturally occurring
CD4CD25 T regulatory cells (Treg cells)(41). Adoptive
transfer of tumor-reactive CD8 T cells plus CD4CD25
Th cells into CD4-deficient hosts induced autoimmunity
and regression of established melanoma. However, transfer of CD4 T cells that contained a mixture of CD4CD25
and CD4CD25 Treg cells or Treg cells alone prevented
effective adoptive immunotherapy. These findings thus
115
suggest that adoptive immunotherapy requires the

absence of naturally occurring CD4CD25 Treg cells to
be effective, and the optimal composition of a cellular agent
should be composed of CD8 cells plus CD4CD25 cells.
ADOPTIVE T CELL IMMUNOTHERAPY OF CANCER

IN LYMPHOPENIC HOST
Studies in the late 1970s demonstrated that the induction
of lymphopenia by sublethal total body irradiation can be
beneficial for the treatment of tumors in mice (42). Chang
et al. reported that the adoptive transfer of immune cells in
the irradiated host confers improved therapeutic effects
compared to the normal host (43). The role of lymphodepletion on the efficacy of T cell therapy is incompletely understood and may depend on the destruction of CD4CD25
regulatory cells, interruption of homeostatic T cell regulation, or abrogation of other normal tolerogenic mechanisms. A report by Dummer et al. indicated that the
reconstitution of the lymphopenic, sublethally irradiated
murine host with syngeneic T cells triggered an antitumor
autoimmune response that required expansion within
lymph nodes (44).
There are several different animal models of lymphopenia that can be utilized. These include the use of whole
body irradiation (WBI), chemotherapy-induced, or genetically altered hosts (i.e., RAG1 knockout mice) that are
deficient of T and B cells. The use of various chemotherapeutic agents to induce lymphopenia would simulate the
clinical setting. Cyclophosphamide (CTX) is an agent that
has been extensively used in murine models and is actively
used in the therapy of certain human cancers. It has been
described to eliminate tumor-induced suppressor cells in
both animal and human settings.
A few years ago, a report described a phase I study of the
adoptive transfer of cloned melanoma antigen-specific T
lymphocytes for therapy of patients with advanced melanoma (45). Clones were derived from peripheral blood
lymphocytes or TILs of patients. Twelve patients received
two cycles of cells. Peripheral blood samples were analyzed
for persistence of transferred cells by TCR-specific PCR.
Transferred cells reached a maximum level at 1 h after
transfer, but rapidly declined to undetectable levels by 2
weeks. The lack of clinical effectiveness of this protocol
suggested that transfer of different or additional cell types,
or that modulation of the recipient host environment was
required for successful therapy. Relevant to these studies
is the clinical experience reported by Rosenberg and coworkers who infused tumor-reactive T cells in melanoma
patients after a nonmyeloablative conditioning regimen
(cyclophosphamide/fludarabine) (46). Conditioning with
the nonmyeloablative chemotherapy before adoptive transfer of activated tumor-reactive T cells enhanced tumor
regression and increased the overall rates of objective clinical responses. Six of thirteen patients demonstrated significant clinical responses as well as autoimmune melanocyte
destruction. In a follow up of this experience in 25 patients,
the conditioning regimen was given prior to adoptive T cell
therapy as before (47). Examination of the T cell persistence through analysis of the specific TCR demonstrated
116
IMMUNOTHERAPY
that there was a significant correlation between tumor

regression and the degree of persistence in peripheral
blood of adoptively transferred T cell clones. Transferred
cells persisted for as long as 2 months in the lymphopenic
setting induced by the conditioning regimen. In contrast,
they presented in the blood for only 2 or 3 weeks without
the prior chemotherapy. These series of studies strongly
suggest that the lymphopenic host induced by the nonmyeloablative conditioning regimen may provide a better
environment for the functioning of the transferred T cells,
and hence improve their therapeutic efficacy. Examination
of the mechanisms involved in the reconstitution of the
lymphodepleted host after adoptive T cell transfer will be
important in identifying methods to improve the efficacy of
T cell therapies for cancer.
REDIRECT EFFECTOR T CELLS TO TUMOR

As mentioned earlier, the low precursor frequency of
tumor-specific T cells in patients hampers routine isolation of these cells for adoptive transfer. To overcome this
problem, targeted adoptive immunotherapy or genetic
adoptive immunotherapy has become an attractive option
for cancer treatment. This strategy can be approached in
two ways: introduction of a chimeric TCR into effector cells;
or introduction of a tumor-specific TCR into nave cells.
The T-body approach uses patient-derived lymphocytes
transfected with chimeric receptor genes constructed with
the variable domains of monoclonal antibodies or cytokines
linked to the constant regions of TCR. The rationale for this
novel approach to redirect effector cells combines the
effector functions of T lymphocytes with the ability of
antibodies or cytokines to recognize predefined surface
antigens or cytokine receptors with high specificity and
in a non-MHC restricted manner.
Eshhar et al. (48) was one of the first to describe this
approach by developing a chimeric receptor gene which
recognized trinitrophenyl (TNP). Retroviral transduction
of the anti-TNP/TCR chimeric gene into a T cell hybridoma
line resulted in gene expression. These gene modified T cells
were cytolytic and released IL-2 in response to TNP-labeled
Daudi cells, but not unlabeled cells. Also among the pioneers
in this area, Hwu et al. (49) developed a recombinant
chimeric receptor against an epitope expressed on the
majority of ovarian cancer cell lines. The TIL were transduced with this chimeric gene and evaluated for immunologic function. The gene modified TIL showed specific lysis of
an ovarian carcinoma cell line, but not nonovarian cell lines.
In a direct comparison, the gene modified TIL showed
greater therapeutic efficacy in vivo than the nontransduced
TIL (49). Pinthus et al. evaluated the therapeutic efficacy of
anti-erbB2 chimeric receptor-bearing human lymphocytes
on human prostate cancer xenografts in a SCID mouse
model (50). Local delivery of erbB2-specific transgenic T
cells to well-established subcutaneous and orthotopic
tumors resulted in retardation of tumor growth and prolongation of animal survival. In a setting of metastatic
cancer (51), anti-erbB2 chimeric receptor-modified T cells
killed breast cancer cells and secreted IFNg in an Ag-specific
manner in vitro. Treatment of established metastatic dis-
ease in lung and liver with these genetically engineered T

cells resulted in dramatic increases in survival of the xenografted mice. In another report, CD4 cells isolated from the
peripheral blood and engrafted with a recombinant immunoreceptor specific for carcinoembryonic Ag (CEA) efficiently lysed target cells in a MHC-independent fashion,
and the efficiency was similar to that of grafted CD8 T cells
(52). In an attempt to further improve the therapeutic utility
of redirected T cells, T lymphocytes were transferred with
CEA-reactive chimeric receptors that incorporate both
CD28 and TCR-zeta signaling domains. T cells expressing
the single-chain variable fragment of Ig (scFv)-CD28-zeta
chimera demonstrated a far greater capacity to control the
growth of CEA xenogeneic and syngeneic colon carcinomas
in vivo compared with scFv-CD28 or scFv-zeta transfected T
cells. This study has illustrated the ability of a chimeric scFv
receptor capable of harnessing the signaling machinery of
both TCR-zeta and CD28 to augment T cell immunity
against tumors (53).
In addition to antibodies, cytokines could also be used to
reconstruct chimeric TCRs. The IL-13 receptor alpha2 (IL13Ra2) is a glioma-restricted cell-surface epitope not otherwise detected within the central nervous system. Kahlon et
al. (54) described a novel approach for targeting glioblastoma multiforme (GBM) with IL-13Ra2-specific CTLs. The
chimeric TCR incorporates IL-13 for selective binding to
IL-13Ra2. This represents a new class of chimeric immunoreceptors that signal through an engineered immune
synapse composed of membrane-tethered cytokine (IL-13)
bound to cell-surface cytokine receptors (IL-13Ra2) on
tumors. Human IL-13-redirected CD8 CTL transfectants
display IL-13Ra2-specific antitumor effector function
including tumor cell cytolysis and cytokine production.
In vivo, the adoptive transfer of genetically modified
CTL clones resulted in the regression of established human
glioblastoma orthotopic xenografts.
The second genetic approach to redirect T cells involves
the introduction of tumor-specific TCRs into nave cells.
Genes encoding tumor antigen-specific TCRs can be introduced into primary human T cells as a potential method of
providing patients with a source of autologous tumor-reactive T cells. Several tumor-associated antigens have been
identified and cloned from human tumors, such as melanoma, breast cancers, and RCC. The antigens have been
identified by their ability to induce T cell reactivity by their
binding to the TCR ab complex. The subsequent cloning of
functional TCR genes capable of recognizing tumor-associated antigens offers a potential opportunity to genetically
modify naive cells that have not been previously exposed to
tumor antigen and to become competent in recognizing
tumor. Cole et al. (55) transfected the cDNA for the TCR a
and b chains of an HLA-A2 restricted, melanoma-reactive
T cell clone into the human Jurkat T cell line. The transfected line was able to mediate recognition of the melanoma
antigen, MART-1, when presented by antigen-presenting
cells. This represented the first report of a naive cellular
construct designed to mediate functional tumor antigen
recognition. A recent study explored the simultaneous
generation of CD8 and CD4 melanoma-reactive T cells
by retroviral-mediated transfer of a TCR specific for HLAA2-restricted epitope of the melanoma antigen tyrosinase
IMMUNOTHERAPY
(56). The TCR-transduced normal human peripheral blood

lymphocytes secreted various cytokines when stimulated
with tyrosinase peptide-loaded antigen-presenting cells or
melanoma cells in an HLA-A2-restricted manner. Rosenberg and co-worker (57) isolated the a and b chains of the
TCR from a highly avid anti-gp100 CTL clone and constructed retroviral vectors to mediate gene transfer into
primary human lymphocytes. The biological activity of
transduced cells was confirmed by cytokine production
following coculture with stimulator cells pulsed with
gp100 peptides, but not with unrelated peptides. The ability of the TCR gene to transfer Ag recognition to engineered
lymphocytes was confirmed by HLA class I-restricted
recognition and lysis of melanoma tumor cell lines. In
addition, nonmelanoma-reactive TIL cultures developed
antimelanoma activity following anti-gp100 TCR gene
transfer. Together, these studies suggest that lymphocytes
genetically engineered to express melanoma antigen-specific TCRs may be of value in the adoptive immunotherapy
of patients with melanoma.
The HPV16 (human papilloma virus type 16) infection of
the genital tract is associated with the development of
cervical cancer in women. The HPV16-derived oncoprotein
E7 is expressed constitutively in these lesions and represents an attractive candidate for T cell mediated adoptive
immunotherapy. In a recent study, Scholten et al. reported
that HPV16E7 TCR gene transfer is feasible as an alternative strategy to generate human HPV16E7-specific T cells
for the treatment of patients suffering from cervical cancer
and other HPV16-induced malignancies (58). These TCR
genes specific for HPV16E7 were isolated and transferred
into peripheral blood-derived CD8 T cells. Biological activity of the transgenic CTL clones was confirmed by lytic
activity and IFNg secretion upon antigen-specific stimulation. Most importantly, the endogenously processed and
HLA-A2 presented HPV16E7 CTL epitope was recognized
by the TCR-transgenic T cells. In a separate study, ovalbumin (OVA)-specific CD4 cells were successfully generated.
Chamoto et al. (59) prepared mouse antigen-specific Th1
cells from nonspecifically activated T cells after retroviral
transfer of TCR genes. These Th1 cells transduced with the
a and b genes of the I-A (d)-restricted OVA-specific TCR
produced IFNg in response to stimulation with OVA peptides or A20 B lymphoma cells expressing OVA as a model
tumor antigen. The TCR-transduced Th1 cells also exhibited
cytotoxicity against tumor cells in an antigen-specific manner. In addition, adoptive transfer of TCR-transduced Th1
cells exhibited potent antitumor activity in vivo.
Genetic alteration of T cells with chimeric receptor
genes or antigen-specific TCR genes confers the redirection
of effector cells to the tumor for its destruction. These
approaches may offer novel opportunities to develop immunocompetent effector cellular reagents and improve the
efficacy of adoptive immunotherapy of cancer.
COMBINED THERAPY
Cancer is a disease that involves multiple gene malfunctions and numerous biochemical and cellular event errors
during its development and metastasis within an indivi-
117
dual. Therefore, it is difficult to achieve success utilizing

adoptive T cell transfer as a monotherapy. The abovereviewed use of vaccination to induce tumor-reactive
pre-effector in vivo; the coadministration of immune adjuvant with T cell transfer; and the gene therapy to redirect T
cells to tumor are all among the strategies taken to elicit
and/or strengthen the efficacy of T cell therapy. Combination therapy is a very common practice during the treatment of diseases. Active vaccine therapy, for example, can
be used in concert with chemotherapy, radiotherapy, or
antibody therapy. Combining a glioma tumor vaccine engineered to express the membrane form of macrophage
colony-stimulating factor with a systemic antiangiogenic
drug-based therapy cured rats bearing 7 day old intracranial gliomas (60). We successfully demonstrated that
local radiotherapy potentiates the therapeutic efficacy of
intratumoral dendritic cell (DC) administration (61),
and that anti-CD137 monoclonal antibody administration
augments the antitumor efficacy of DC-based vaccines
(62).
In order to enhance the efficiency of T cell therapy,
various strategies have been employed accompanying cell
transfer. These combined therapies include cell transfer in
combination with intratumoral expression of lymphotactin
(63), DC vaccination (64), or blockade of certain molecules
expressed in tumor cells, such as B7-H1 (65).
One of the major obstacles to successful adoptive T cell
therapy is the lack of efficient T cell infiltration of tumor.
Combined intratumoral lymphotactin (Lptn) gene transfer
into SP2/0 myeloma tumors and adoptive immunotherapy
with tumor specific T cells eradicated well-established
SP2/0 tumors in six of eight mice, and dramatically slowed
down tumor growth in the other two mice (63). Cell tracking using labeled T cells revealed that T cells infiltrated
better into the Lptn-expressing tumors than non-Lptnexpressing ones. These data provide solid evidence of a
potent synergy between adoptive T cell therapy and Lptn
gene therapy as a result of facilitated T cell targeting.
Dendritic cells are well-known potent antigen-presenting
cells. Hwu and co-workers (64) reported that DC vaccination could improve the efficacy of adoptively transferred T
cells to induce an enhanced antitumor immune response.
Mice bearing B16 melanoma tumors expressing the gp100
tumor antigen were treated with activated T cells transgenic for a TCR specifically recognizing gp100, with or
without concurrent peptide-pulsed DC vaccination. Antigen-specific DC vaccination induced cytokine production,
enhanced cell proliferation, and increased tumor infiltration of adoptively transferred T cells. The combination of
DC vaccination and adoptive T cell transfer led to a more
robust antitumor response than the use of each treatment
individually. This work shows that in addition to their
ability to initiate cell-mediated immune responses by
stimulating naive T cells, dendritic cells can strongly boost
the antitumor activity of activated T cells in vivo during
adoptive immunotherapy. Certain cell surface molecules,
expressed either on tumor cells or on T cells, have demonstrated have demonstrated potential suppressive impact
on the adoptive T cell immunotherapy. For example, during the last few years, new members of the B7 family
molecules have been identified, for example, B7-H1, which
118
IMMUNOTHERAPY
is constitutively expressed on 66% of freshly isolated squamous cell carcinomas of the head and neck (SCCHN) (65).
When B7-H1-negative mouse SCC line, SCCVII, was transfected to express B7-H1, all of the animals succumbed to
B7-H1/SCCVII tumors even after adoptive T cell immunotherapy. However, the infusion of B7-H1 blocking monoclonal antibody with activated T cells cured 60% of
animals. The data support B7-H1 blockade as a new
approach to enhance the efficacy of T cell immunotherapy.
These findings also illuminate a new potential application
for the blockade of certain negative costimulation molecules on T cells, for example, CTLA-4 and programmed
death-1 (PD-1) molecules. This kind of blocking may augment the therapeutic efficacy mediated by the transferred
T cells. The blockade can be done using specific monoclonal
antibodies, soluble ligands for CTLA-4 or PD-1, or by
synthesized antagonists. In addition, effector cells can be
derived from the animals deficient in the relevant molecules for preclinical investigations.
Immune tolerance of tumor-bearing host represents
another major obstacle for the successful use of adoptive
T cell immunotherapy. A recent study examined the
requirement for assistance to the low affinity tumor-specific CD8 T cells transferred into tumor-bearing mice
(66). The TCR transgenic mice expressing a class Irestricted hemagglutinin (HA)-specific TCR (clone 1
TCR) were generated. Upon transfer into recipient mice
in which HA is expressed at high concentrations as a
tumor-associated Ag, the clone 1 TCR CD8 T cells exhibited very weak effector function and were soon tolerized.
However, when HA-specific CD4 helper cells were cotransferred with clone 1 cells and the recipients were
vaccinated with influenza, clone 1 cells were found to
exert a significant level of effector function and delayed
tumor growth. This work shows that in order to optimize
the function of low avidity tumor-specific T cells after
adoptive transfer, additional measures need to be taken
to help break the host tolerance.
Effective tumor therapy requires a proinflammatory
microenvironment that permits T cells to extravasate
and to destroy the tumor. Proinflammatory environment
can be induced by various chemical, physical, and immunological protocols. Greater extent of success can be
expected by combining adoptive T cell therapy with the
traditional cancer treatment methods, for example, surgery, chemotherapy, and radiation therapy, as well as with
different forms of immunotherapeutic strategies, such as
vaccine, antibody, cytokines, gene therapy, and so on. The
factors to be combined can involve two or more approaches.
In summary, adoptive immunotherapy utilizing tumorreactive T cells offers a promising alternative approach for
the management of cancer. Through the endeavors of
clinical and basic research scientists during the last two
decades, the process of adoptive T cell therapy of cancer has
evolved from its original single-step approach into its
current multiple-step procedure. Successful T cell immunotherapy of cancer is the outcome of this multi-step
process that depends on successful Ag priming, numerical
amplification of low frequency Ag-specific precursors, use
of immune adjuvants, and efficient infiltration of tumors in
all metastatic sites by effector T cells. New directions in
this field include the identification and application of

tumor-reactive subpopulation of T cells, creation of a lymphopenic environment in the recipient host, and the redirection of the effector cells toward the tumor. Development
of these latter techniques and the combined use of different
therapeutic strategies may further improve the efficacy of
the immunotherapy of human cancer employing adoptive T
cell transfer. Studies and developments of immunotherapy
for cancer should accelerate the application of this strategy
in infectious disease, autoimmune disease and other disease managements.
BIBLIOGRAPHY
1. Chang AE, et al. Adoptive Immunotherapy with vaccineprimed lymph node cells secondarily activated with antiCD3 and interleukin-2. J Clin Oncol 1997;15:796.
2. Chang AE., et al. Generation of vaccine-primed lymphocytes
for the treatment of head and neck cancer. Head and Neck
2003;25:198.
3. Chang AE, et al. Phase II trial of autologous tumor vaccination, Anti-CD3-activated vaccine-primed lymphocytes, and
Interleukin-2 in stage IV renal cell cancer. J Clin Oncol
2003;21:884.
4. Shu S, Chou T, Rosenberg SA. Generation from tumor-bearing
mice of lymphoid cells with in vivo therapeutic efficacy. J
Immunol 1987;139:295304.
5. Chou T, Chang AE, Shu S. Generation of therapeutic T lymphocytes from tumor-bearing mice by in vitro sensitization:
Culture requirements and characterization of immunologic
specificity. J Immunol 1988;140:24532461.
6. Geiger J, Wagner P, Shu S, Chang AE. A novel role for
autologous tumor cell vaccination in the immunotherapy of
the poorly immunogenic B16-BL6 melanoma. Surg Oncol
1992;1:199208.
7. Geiger J, et al. Generation of T- cells reactive to the poorly
immunogenic B16-BL6 melanoma with efficacy in the treatment of spontaneous metastases. J Immunother 1993;13:153
65.
8. Li Q, et al. Immunological effects of BCG as an adjuvant in
autologous tumor vaccines. Clin Immunol 2000;94:6472.
9. Restifo N, et al. A nonimmunogenic sarcoma induced with the
cDNA for interferon-g elicits CD8 T cells against the wildtype tumor: Correlation with antigen presentation capability.
J Exp Med 1992;175:14231431.
10. Chang AE, et al. Immunogenetic therapy of human melanoma
utilizing autologous tumor cells transduced to secrete GM
CSF. Hum Gene Ther 2000;11:839850.
11. Liu J, et al. Ex vivo activation of tumor-draining lymph node T
cells reverses defects in signal transduction molecules. Can
Immunol Immunother 1998;46:268.
12. Yoshizawa H, Chang AE, Shu S. Specific adoptive immunotherapy mediated by tumor-draining lymph node cells
sequentially activated with anti-CD3 and IL-2. J Immunol
1991;147:72937.
13. Li Q, Furman SA, Bradford CR, Chang AE. Expanded tumorreactive CD4 T-Cell responses to human cancers induced by
secondary anti-CD3/anti- CD28 activation. Clin. Can. Res.
1999;5:461.
14. Li Q, et al. Therapeutic effects of tumor-reactive CD4 cells
generated from tumor-primed lymph nodes using anti-CD3/
anti-CD28 monoclonal antibodies. J Immunother 2002;25:304.
15. Ito F, et al. Antitumor therapy reactivity of Anti-CD3/AntiCD28 bead-activated lymphoid cells: Implications for cell in a
murine model. J Immunother 2003;26:222.
IMMUNOTHERAPY
16. Li Q, et al. Polarization effects of 4-1BB during CD28 costimulation in generating tumor-reactive T Cells for cancer
immunotherapy. Can. Res. 2003;63:2546.
17. Aruga A, et al. Type 1 versus type 2 cytokine release by Vb T
cell subpopulations determines in vivo antitumor reactivity:
IL-10 mediates a suppressive role. J Immunol 1997;159:664
673.
18. Kagamu H, Shu S. Purification of L-selectinlow cells promotes
the generation of highly potent CD4 antitumor effector T
lymphocytes. J Immunol; 1998;160:34443452.
19. Hart-Meyers J, et al. Cutting Edge: CD94/NKG2 is expressed
on Th1 but not Th2 cells and costimulates Th1 effector functions. J Immunol 2002;169:53825386.
20. Hou W, et al. Pertussis toxin enhances Th1 responses by
stimulation of dendritic cells. J Immunol 2003;170:1728
1736.
21. Li Q, et al. Synergistic effects of IL-12 and IL-18 in skewing
tumor-reactive T-cell responses towards a type 1 pattern. Can.
Res. 2005;65:1063.
22. Topalian SL. et al. Immunotherapy of patients with advanced
cancer using tumor-infiltrating lymphocytes and recombinant
interleukin-2: a pilot study. J. Clin. Oncol. 1988;6:839.
23. Shrikant P, Mescher MF. Opposing effects of IL-2 in tumor
immunotherapy: promoting CD8 T cell growth and inducing
apoptosis. J. Immunol. 2002;169:1753.
24. Rosenberg SA, et al. Experience with the use of high-dose
IL-2 in the treatment of 652 cancer patients. Ann Surg
1989;210:474.
25. Roychowdhury S, et al. Failed adoptive immunotherapy with
tumor-specific T cells: Reversal with low-dose interleukin 15
but not low-dose interleukin 2. Can Res 2004;64:8062.
26. Nacsa J, et al. Contrasting effects of low-dose IL-2 on vaccineboosted Simian Immunodeficiency Virus (SIV)-specific CD4
and CD8 T cells in macaques chronically infected in SIVmac251. J Immunol 2005;174:1913.
27. Poggi A, et al. Tumor-induced apoptosis of human IL-2activated NK cells: role of natural cytotoxicity receptors. J
Immunol 2005;174:2653.
28. Yajima T, et al. A novel role of IL-15 in early activation of
memory CD8 CTL after reinfection. J Immunol 2005;174:
35903597.
29. Kamiya S, et al. An indispensable role for STAT1 in IL-27induced T-bet expression but not proliferation of nave CD4 T
cells. J Immunol 2004;173:38713877.
30. Osaki T, et al. IFN-g-inducing factor/IL-18 administration
mediates IFN-g-and IL-12-independent antitumor effects. J
Immunol 1998;160:1742.
31. Palmer DC, et al. Vaccine-stimulated adoptively transferred
CD8 T cells traffic indiscriminately and ubiquitously while
mediating specific tumor destruction. J Immunol 2004;173:
72097216.
32. Mukai S, Kjaergaard J, Shu S, Plautz GE. Infiltration of
tumors by systemically transferred tumor-reactive T lymphocytes is required for antitumor efficacy. Can Res 59: 5245.
33. Skitzki J, et al. Donor cell cycling, trafficking, and accumulation during adoptive immunotherapy for murine lung metastases. Can Res 2004;64:2183.
34. Meidenbauer N, et al. Survival and tumor localization of
adoptively transferred melan-A-specific T cells in melanoma
patients. J Immunol 2003;170:21612169.
35. Chi H, Flavell RA. Cutting Edge: Regulation of T cell trafficking and primary immune responses by sphingosine 1-phosphate receptor 1. J Immunol 2005;174:24852488.
36. Wang LX, Chen BG, Plautz GE. Adoptive immunotherapy of
advanced tumors with CD62 L-selectinlow tumor-sensitized T
lymphocytes following ex vivo hyperexpansion. J Immunol
2002;169:33143320.
119
37. Peng L, et al. Tumor-induced L-selectinhigh suppressor T cells

mediate potent effector T cell blockade and cause failure of
otherwise curative adoptive immunotherapy. J Immunol
2002;169:48114821.
38. Tanigawa K, et al. Tumor-specific responses in lymph nodes
draining murine sarcomas are concentrated in cells expressing
P-selectin binding sites. J Immunol 2001;167:30893098.
39. Xu H, et al. Recruitment of IFN-g-producing (Th1-like) cells
into the inflamed retina in vivo is preferentially regulated by
P-selectin glycoprotein ligand 1:P/E-selectin interactions. J
Immunol 2004;172:32153224.
40. Shu S, et al. Stimulation of tumor-draining lymph node cells
with superantigenic staphyloccocal toxins leads to the generation of tumor-specific effector cells. J Immunol 1994;152:1277
1288.
41. Antony PA, et al. CD8 T cell immunity against a tumor/selfantigen is augmented by CD4 T helper cells and hindered by
naturally occurring T regulatory cells. J Immunol 2005;174:
25912601.
42. Mule JJ, Jones FR, Hellstrom I, Hellstrom KE. Selective
localization of radio-labeled immune lymphocytes into syngeneic tumors. J Immunol 1979;123:600.
43. Chang AE, et al. Differences in the effects of host suppression
on the adoptive immunotherapy of subcutaneous and visceral
tumors. Can Res 1986;46:3426.
44. Dummer W, et al. T cell homeostatic proliferation elicits
effective antitumor autoimmunity. J Clin Inves 2002;110:185.
45. Dudley ME, et al. Adoptive transfer of cloned melanomareactive T lymphocytes for the treatment of patients with
metastatic melanoma. J Immunoth 2001;24:363373.
46. Dudley ME, et al. Cancer regression and autoimmunity in
patients after clonal repopulation with antitumor lymphocytes. Science 2002;298:850.
47. Robbins PF, et al. Cutting Edge: Persistence of transferred
lymphocyte clonotypes correlates with cancer regression in
patients receiving cell transfer therapy. J Immunol 2004;173:
71257130.
48. Eshhar Z, Waks T, Schindler DG, Gross G. Specific activation
and targeting of cytotoxic lymphocytes through chimeric single chains consisting of antibody binding domains and the g or
z subunits of the immunoglobulin and T cell receptors. Proc
Natl Acad Sci USA 1993;90:720724.
49. Hwu P, et al. In vivo antitumor activity of T cells redirected
with chimeric antibody/T cell receptor genes. Can Res
1995;55:33693373.
50. Pinthus JH, et al. Immuno-gene therapy of established prostate tumors using chimeric receptor-redirected human lymphocytes. Can Res 2003;63:24702476.
51. Kershaw MH, et al. Gene-engineered T cells as a superior
adjuvant therapy for metastatic cancer. J Immunol 2004;173:
21432150.
52. Hombach A, et al. CD4 T cells engrafted with a recombinant
immunoreceptor efficiently lyse target cells in a MHC antigenand-fas independent fashion. J Immunol 2001;167:10901096.
53. Haynes NM, et al. Rejection of syngeneic colon carcinoma by
CTLs expressing single-chain antibody receptors codelivering
CD28 costimulation. J Immunol 2002;169:57805786.
54. Kahlon KS, et al. Specific recognition and killing of glioblastoma multiforme by interleukin 13-zetakine redirected cytolytic T cells. Can Res 2004;64:91609166.
55. Cole DJ, et al. Characterization of the functional specificity of
a cloned T cell receptor heterodimer recognizing the MART-1
melanoma antigen. Can Res 1995;55:748752.
56. Roszkowski JJ, et al. Simultaneous generation of CD8 and
CD4 melanoma-Reactive T cells by retroviral-mediated
transfer of a single T-cell receptor. Can Res 2005;65:1570
1576.
120
57. Morgan RA, et al. High efficiency TCR gene transfer into
primary human lymphocytes affords avid recognition of melanoma tumor antigen glycoprotein 100 and does not alter the
recognition of autologous melanoma antigens. J Immunol
2003;171:32873295.
58. Scholten KB, et al. Preservation and redirection of HPV16E7specific T cell receptors for immunotherapy of cervical cancer.
Clin Immunol 2005;114:119129.
59. Chamoto K, et al. Potentiation of tumor eradication by adoptive immunotherapy with T-cellreceptor gene-transduced Thelper type 1 cells. Can Res 2004;64:386390.
60. Jeffes EWB, et al. Antiangiogenic drugs synergize with a
membrane macrophage colony-stimulating factory-based
tumor vaccine to therapeutically treat rats with an established
malignant intracranial glioma. J Immunol 2005;174:2533
2543.
61. Teitz-Tennenbaum S. et al. Radiotherapy potentiates the
therapeutic efficacy of intratumoral dendritic cell administration. Can Res 2003;63:84668475.
62. Ito F, et al. Anti-CD137 monoclonal antibody administration
augments the antitumor efficacy of dendritic cell-based vaccines. Can Res 2004;64:8411.
63. Huang H, Li F, Gordon JR, Xiang J. Synergistic enhancement
of antitumor immunity with adoptively transferred tumorspecific CD4 and CD8 T cells and intratumoral lymphotactin transgene expression. Can Res 2002;62:2043.
64. Lou Y, et al. Dendritic cells strongly boost the antitumor
activity of adoptively transferred T cells in vivo. Can Res
2004;64:6783.
65. Strome SE. B7-H1 blockade augments adoptive T-cell immunotherapy for squamous cell carcinoma. Can Res 2003;63:
6501.
66. Lyman MA, et al. The fate of low affinity tumor-specific CD8
T cells in tumor-bearing mice. J Immunol 2005;174:2563
2572.
See also BORON
NEUTRON CAPTURE THERAPY; MONOCLONAL ANTIBODIES.
HELMUT HUTTEN
University of Technology
Graz, Australia
INTRODUCTION
Plethysmography is a volumetric method, that is, a method
for the assessment of a volume (the Greek words plethys
and plethora mean full and fullness, respectively). Impedance plethysmography is based on the measurement of
passive electrical properties of biological tissues. Those
passive electrical properties are parameters of the so-called
bioimpedance. The first publication about impedance
plethysmography by Nyboer et al. (1) dates back to 1943.
Pioneering contributions to the basic understanding of the
relations between the assessment of volumes by impedance
plethymosgraphy and the electrical properties of biological
tissue have been provided by Schwan et al. (2) already in
1955. But already by the end of the nineteenth century
Stewart had used the recording of electrical conductivity to
study transit times between different sites of the body after
injection of saline into the circulation (3). Blood flow record-
ing is one of the most relevant fields for the clinical application of impedance plethysmography nowadays.
Impedance plethysmography is a volumetric method
that aims to assess a volume or changes of a volume.
Usually, a volume is the filling volume of a space that is
enclosed by geometric boundaries. In this case, volumetry
means the determination of the boundaries with subsequent assessment of the volume within the boundaries.
Those boundaries can be determined by the impedance
method if the electrical properties of the substances on
both sides of the boundaries are different.
Impedance plethysmography, however, can also be
applied to the assessment of volumes that are not lumped
compartments within geometric boundaries, for example,
it can be used for the volumetric measurement of a certain
component within a mixture. Such components may be
cells (e.g., the volume of cells in blood), tissues (e.g., the
volume of fat tissue in the body), spaces with different
composition (e.g., intra- and extracellular spaces), or the
volume of the air that is enclosed in the alveoli of lung
tissue. In that case, volumetry means the estimation of the
space that would be occupied by the respective component
if it would be concentrated in one single lumped compartment. Usually, this volume is estimated as a percentage of
the whole distribution volume, for example, the volume of
cells in blood or the content of water in the whole body. The
electrical properties of the respective component must be
different from those of all other components. The volumetric assessment does not require a homogeneous distribution of the considered component within the given
space if the actual distribution can be taken into account,
for example, by a model. Under certain conditions, a tissue
can be identified by specific features like morphological
structure and/or chemical composition if those features are
related with its electrical properties.
The typical application of plethysmography in clinical
routine is the diagnosis of those diseases for which the
measurement of volumes or changes of volume renders
possible the interpretation of functional disorders or functional parameters. The most widely and routinely applied
diagnostic examinations are concerned with:
1. Heart: Cardiac mechanical disorders by impedance
cardiography (i.e., pumping insufficiency by measuring cardiac stroke volume, including heart rate and
other cardiac parameters like ejection period). This
application is discussed in another article.
2. Peripheral circulation: Vascular disorders by impedance rheography (i.e., deep venous thrombosis by
impedance phlebography, and estimation of blood
flow in the brain or other peripheral vessels).
3. Lung: Ventilatory disorders by impedance pneumography (i.e., insufficient ventilation by monitoring
the tidal volume and/or respiratory rate).
METHODOLOGY
Impedance plethysmography is a noninvasive method that
employs contacting, usually disposable electrodes, in most
cases metal-gel electrodes, for example, with Ag/AgCl for
the metal plate. Usually, the electrodes have circular

geometry; however, other geometries might be preferable,
for example, band-like geometry for segmental measurement at the extremities. It must be considered that the
metal plates of electrodes are areas with the same potential, and therefore may affect the electromagnetic field
distribution in the considered object. Electrodes with small
areas help to reduce that effect, whereas electrodes with
large areas render it possible to reach a more homogenous
current field in the measured object.
Contacting electrodes can easily be attached to the skin
or surface of the measurement object, usually by an adhesive material that is already fixed to the electrode. Only in
special cases, for example, for research purposes, does the
measurement require invasive application.
Different contactless measurement modes gain increasing attention, for example,
1. Microwave-based methods with antennas as applicators and measurement of the scattered electromagnetic field.
2. Methods based on exploiting the magnetic instead of
the electrical properties: inductive plethysmography
that uses coils and records the changes in the inductance and magnetic susceptibility plethysmography
that employs strong magnetic fields and records the
changes in the magnetic flux, for example, by superconducting quantum interference devices (SQUID).
All bioimpedance-based methods are aiming at recording either the effect on the applied electromagnetic field by
the measurement object or the response of the measurement object to the application of the electromagnetic field.
The directly measured quantities are electrical quantities,
for example, voltages or currents. Those quantities are
actually imaging electrical coefficients, for example, conductivity, permittivity, or resistivity, which are materialspecific parameters of the tissue impedance and monitor its
morphological structure and/or chemical composition.
With these material-specific parameters the geometric
boundaries are determined and used for the estimation
of the volume or volume changes. Relations between the
measured electrical parameters and the volume are
usually based on models. The employed models can be very
simple, for example, described by simple geometric boundaries like cylinders, spheres, or ellipsoids. More complex
3D models may be described by the finite element method
(FEM) or similar approaches, which allows simulating the
distribution of the electromagnetic field in the measured
object, that is, the current pathways and the iso-potential
planes. Sophisticated iterative optimization procedures are
employed to match the simulated values with the measured ones (4).
Electrical impedance tomography (EIT) is a direct
approach for determining the 3D geometry of those compartments with the same material-specific parameters in a
biological object like the human torso or an extremity. This
technique uses multiple electrodes, usually arranged in a
plane. Mapping (or imaging) of the impedance distribution
in the examined cross-sectional layer requires the solution
of the inverse or back-projection problem. Actually, the
121
obtained 2D image is a pseudo-3D image since the current

pathways are not constrained to the examined layer. Electrical impedance tomography supplies a comparable nearanatomic cross-sectional image comparable to those of
other CT-based procedures [e.g., X-ray CT, NMR, or ultrasound (US)], however, with very poor spatial resolution.
Boundaries of compartments with the same material-specific coefficients are found by segmentation. Segmented
areas with assumed thickness of the single layers are used
for volume estimation. Changes in the volume can be
assessed by comparing the volumes obtained in consecutive
images.
It is a characteristic feature of all methods that record
the electrical bioimpedance that the evoked response
depends on the strength of the local electromagnetic field.
For this reason, it has to be taken into account that the
resulting current density may be inhomogeneous in the
considered tissue volume. Causes for such nonhomogeneity
may be geometric constraints (e.g., a nonregular shape of
the considered volume); the composition of the tissue
within the considered volume that may be a mixture of
tissues with different electrical properties (e.g., blood with
low resistivity, or bone with high resistivity, as compared
with skeletal muscle). Those different tissues may electrically be switched in parallel or serial order; and the size
and the location of the current-feeding electrodes. The
current density is higher in regions near to the feeding
electrodes than in distant regions. Consequently, the
regions near to the feeding electrodes give the strongest
contribution to the measured voltage.
This requires (1) careful selection of the current-feeding
site. In the tetrapolar mode also the position of the voltagesensing electrodes must be taken into account; and (2)
appropriate consideration of the inhomogeneous distribution of the electrical parameters within the considered
tissue volume, that is, the course of blood vessels.
Special electrode arrangements have been developed for
certain applications in order to minimize the measurement
errors. Concentric multielectrode arrangements with the
outer electrodes on a potential different from that of the
inner electrode have been proposed with the objective to
optimize the current distribution in the measured volume.
The frequency that can be used for the measurement of
the passive electrical properties of biological tissue ranges
from very low frequencies to some gigahertz. The most
popular frequency band for impedance plethysmography is
between 1 kHz and 10 MHz. This frequency band encloses
the so-called b-dispersion, which is actually a dielectric or
structural relaxation process. The b-dispersion is also
known as MaxwellWagner relaxation. It is characterized
by a transition in the magnitude of the electrical parameters with frequency. This transition is caused by the fact
that cellular membranes have high impedance below and
low impedance above that b-dispersion. For frequencies
distinctly below the b-dispersion, the current flow is
restricted to the extracellular space. For frequencies distinctly above the b-dispersion, the current can pass
through the cellular membrane. Consequently, with frequencies distinctly below the b-dispersion only the volume
or volume changes of the extracellular space will be monitored, whereas with frequencies distinctly above the
122
Table 1. Compilation of Typical Values of Resistivity (Vm) of Various Body Tissuesa

Frequency
10 Hz
Muscle, skeletal
Muscle, heart
Liver
Kidney
Brain
Fatty tissue
Blood
9.6
9.6
10.0
100 Hz
1 kHz
10 kHz
100 kHz
1 MHz
10 MHz
100 MHz
8.8
9.3
8.7
8.1
8.0
8.6
7.6
6.0
7.6
6.1
23.2
1.6
2.0
2.1
4.6
1.9
6.0
1.8
2.0
2.8
1.8
5.3
1.6
1.6
2.8
1.4
3.7
1.4
1.5
1.7
1.3
1.5
1.5
1.5
1.4
0.9
0.8
Note that those values must not be assumed to represent exact figures since they do not consider important details like species, preparation of sample, time
after excision, temperature, or the procedure and protocol for their measurement. The values are compiled from many different sources and, if necessary
transformed to resistivity.
b-dispersion, the total volume (i.e., both extra- and intracellular space) or changes of this total volume can be
recorded. Using at least one frequency below and another
one above the b-dispersion allows determining the ratio of
extra- and intracellular spaces, and hence also fluid shifts
between these spaces.
Special applications of this approach are the monitoring
of fluid exchange processes during hemodialysis (5) and
orthostatic challenges (6), the control and management of
fluid infusion therapy, the detection of lung edema, and the
viability surveillance of organs after blood flow has been
stopped during surgery or when the organs are preserved
for transplantation (7,8). The viability surveillance is
based on the fact that oxygen deficiency with the subsequent lack of energy-rich substrates causes a failure of the
active transmembraneous ionic transport mechanisms
and, as a consequence, leads to an intracellular edema
(i.e., an increase of the intracellular volume). This
approach has also been investigated for graft rejection
monitoring.
The passive electrical properties are specific for each
tissue. They are mainly depending on the content of water,
the ratio of extra- and intracellular space, the concentration of electrolytes, and the shape of the cells and their
orientation in the electrical field (e.g., of the fibers of
skeletal and cardiac muscle). Table 1 shows a compilation
of typical values of resistivity of various body tissues. It
must be taken into account, however, that these values do
not represent exact figures. Exact figures need detailed
information about species, preparation of the sample, time
after excision, measurement temperature, the employed
method, and the protocol for the measurement. Comprehensive data compilations with the supplement of those
details are found in Refs. 9 and 10.
These tissue-specific properties can be used for special
applications, such as the analysis of the tissue composition
or for tissue characterization by Impedance Spectroscopy.
Those methods are the subject of another article and will
not be discussed here in detail. A very popular application
is the determination of total body water (11) or of whole
body composition, for example, the determination of the
percentage of body fat in order to support adequate nutrition management or control of physical exercises. Such
approaches aim for the estimation of the compartmental
volume of a certain tissue (e.g., fat) that is mixed with
another tissue (e.g. fat-free tissue) in a common space (i.e.,

the body or an extremity).
FUNDAMENTALS OF BIOIMPEDANCE MEASUREMENT
The most important principle for bioimpedance measurements is the adequate modeling of the passive electrical
behavior of the tissue by an equivalent electrical circuit.
The validity of simple models is restricted to narrow frequency ranges (e.g., the b-dispersion) and/or to simple
geometric shapes of the biological object (e.g., cylinders
as a model for extremities). The most widely accepted
models for the bioimpedance in the frequency range around
the b- dispersion are the RC-networks shown in Fig. 1.
These models represent the spatially distributed electrical
properties by discrete components. Actually, they are only
simplified 2D models. The network shown in Fig. 1a is
mimicking the biological system and its histological structure. It represents both the extracellular and intracellular
space by the resistors Re and Ri, respectively, and the cell
membrane by the capacitor Cm. Since the current passes
twice the membrane when flowing through the cell, the two
capacitors Cm* in series with Ri can equivalently be
expressed by one single capacitor Cm in series with Ri.
This network is usually replaced by the one shown in Fig.
1b in which Rs is arranged in series with the parallel circuit
of Rp and Cp. These components have no relation with real
histological structures. The parameter Rs corresponds to
the parallel circuitry of Re and Ri as can be demonstrated
for high frequencies. The parameter Rs can be considered to
be very small as compared with Rp. In many cases, RS may
Figure 1. RC-networks modeling tissue impedance. The model in

(a) mimicks morphological structures, whereas the model in (b)
shows the electrically equivalent, but, more usual circuitry.
(a)
Sine wave signal

generator
Measuring
device
Sine wave signal

generator
(b)
123
Measuring
device
FE
Electrode
SE
FE
Skin
Tissue
Current
pathways
Current
pathways
FE...Feeding electrode
SE...Sensing electrode
even be neglected for cases of simplification, that is, the

electrical model is simply a parallel circuit of a resistor and
a capacitor (e.g., RpkCp).
When using contacting electrodes, different approaches
are possible for the measurement of the bioimpedance. The
most important feature is the number of employed electrodes, usually two or four electrodes. More than 4 electrodes,
up to 128 electrodes, are primarily used for CT-based
impedance imaging like EIT. The two-electrode configuration is called the bipolar mode, and the four-electrode
configuration is the tetrapolar mode (Fig. 2). The bipolar
mode can be compared with the usual method for impedance measurement by feeding a current into the measurement object and recording the voltage or vice versa. In the
tetrapolar mode, two electrodes are the feeding electrodes
(usually the outer electrodes) and the other two electrodes
are the sensing electrodes (usually the inner ones). In the
tetrapolar mode, more than two sensing electrodes can be
employed, for example, if monitoring of serial segments at
the extremities are to be achieved.
The interface between the electrode with the metallic
plate on the one side and the electrolyte on the other side is
the boundary where a current carried by electrons is
transformed into a current carried by ions. The electrolyte
may either be contained in the gel of the electrodes or be the
electrolytic fluid in the tissue. The basic process of the
charge transfer from electrons to ions is a chemical reaction
(12). The simplest model of such an interface is again an
impedance consisting of a parallel circuit with a resistor RF
(the Faraday resistance) and a capacitor CH (the Helmholtz
capacitance), i.e., RFkCH (Fig. 3b). Real electrodes show a
polarization effect that is caused by a double layer of
opposite charges at the interface, actually the Helmholtz
capacitance (Fig. 3a). Therefore, the electrode model with
RFkCH has to be supplemented with an additional voltage
source EP. The steady-state condition is reached if the
tendency of metallic ions to enter the electrolyte and leave
behind free electrons is balanced by the electrostatic voltage originating from the double layer. After disturbances,
for example, by charge transfer forced by an externally
applied voltage, another equilibrium for the double-layer
voltage is reached with a time constant depending on RF
and CH. All these components may have poor stability with
time, especially in the period immediately after attaching
the electrode on the skin. For surface electrodes, it must
also be taken into account that the impedance of the skin,
especially the stratum corneum, which is the outmost
Figure 2. Bioimpedance measurement set-up. (a)

Shows the two-electrode configuration, whereas
the four-electrode configuration is depicted in (b).
epidermal layer, can be much larger than the impedance

of the deeper tissue (e.g., skeletal muscle), which is in a
complex parallel-serial arrangement with the skin. Sweating underneath the electrode lowers the electrode-tissue
transimpedance. For the measurement of the impedance of
deeper tissues the adequate preparation of the skin by
abrasion, stripping, or puncturing at the site of the electrodes might be necessary in order to diminish the transimpedance. This transimpedance depends on the size of
the electrode (i.e., the contacting area) and the measurement frequency, and . . . additionally on the pressure with
which the electrode is attached to the skin. For electrodes, a
typical value for the transimpedance is 100200 Vcm2.
In the bipolar mode, the two electrodeelectrolyte interfaces are in series with the actual bioimpedance of the
measured object. Therefore, the recorded impedance is
always the sum of at least three impedances. The impedance of the biological sample cannot be calculated as an
individual quantity from the recorded impedance value.
This is the most serious shortcoming of the bipolar mode.
In the tetrapolar mode, the electrodeelectrolyte interface usually can be neglected if the measurement is performed with a device with high input impedance (i.e., with
very low current passing across the electrodetissue inter+
Me
e
e
Me
Me
+
+
Me
(a)
Metal electrode
Electrolyte
RF
(b)
CH
Figure 3. Metal electrode: electrolyte interface. (a) Illustrates

the steady-state condition at the boundary. Following the
tendency of metallic ions (Me) to enter the electrolyte and to
leave behind free electrons, the steady state is reached when this
tendency is balanced by the electrostatic voltage originating from
the double layer. (b) Shows the simplified model with only the
passive electrical components, that is, the Faraday resistance RF
and the Helmholtz capacitance CH.
124
face). A drawback of the tetrapolar mode, however, is that

the measured impedance value cannot be exactly assigned
with a certain volume of the tissue between the sensing
electrodes, even if the sensing electrodes are positioned on
a straight line connecting the two feeding electrodes. Band
electrodes that are attached at the whole circumference
(e.g., at the extremities), yield more valid results. If circular electrodes are applied and both the feeding and the
sensing electrodes are placed on the circumference of a
cross-section (e.g., around the thorax or the head), it is
nearly impossible to assign the actually measured impedance value with a certain volume within that cross-section
due to the complex current field. In those cases, the monitored volume consists of a serial-parallel circuitry of different tissue layers with different electrical properties
(e.g., skin, subcutaneous soft tissue, bone, deeper tissues).
For this reason, the result of ventilation measurements
with electrodes, either disk or band electrodes, placed on
the thorax may be affected by the higher conductivity of
extra-thoracic muscles as compared with the very low
conductivity of the rib cage, which prevents the entrance
of current into the pulmonary tissue that is actually the
object of interest. Sweating may additionally cause another
low impedance parallel circuit along the skin and, thus,
yield considerable measurement errors. The situation is
similar for the measurement of brain parameters, for
example, brainblood flow or brain edema, with electrodes
placed on the scalp. Since the conductivity of the extracranial soft tissue (e.g., skin, muscle) is much higher than
the conductivity of the bony skull, only few current pathways will pass through the brain.
The different instrumental approaches for measuring
the bioimpedance in the frequency range of the b-dispersion are the impedance bridge (e.g., an extension of the
classical Wheatstone bridge); the self-balanced active
bridge; the resonance method that is mainly a compensation method; the pulse method that is a time domain
procedure and not widely used; the voltage-current
method, based either on feeding a constant voltage (i.e.,
from a generator with low source impedance) and monitoring the resulting current or on feeding a constant current
(i.e., from a generator with a sufficiently high source
impedance, 100 kV since the load impedance may be
up to 1 kV) and monitoring the resulting voltage. If
employing the bipolar mode, it would be more correct in
this case to use the term transimpedance than impedance
for the actually measured quantity between the electrodes.
For the tetrapolar configuration only the voltage-current method is applicable. Phase angle measurements
employ an ohmic resistor in series with the measuring
object as reference. Absolute phase angle measurements
are questionable even for the bipolar configuration since
the measured phase angle always includes the phase shifts
that are caused by the two electrodeskin contacts and
depend on the actual values of the Faraday resistance and
the Helmholtz capacitance. If the Faraday resistance is
small and the Helmholtz capacitance is fairly large, the
phase shift by the electrodeskin interface may become
negligible. This is one of the advantages of electrodes with
artificially increased surfaces, for example, electrodes with
porous or fractally coated surfaces that might be obtained
by sputtering or chemical processes, as compared with

polished surfaces.
Usually, the measurement is performed with a constant-voltage generator for technical reasons. The applied
feeding signal, whether voltage or current, should be sinusoidal with a very low distortion factor (i.e., with a low
content of harmonics) and with high stability both in
amplitude and frequency. Any modulation of the feeding
signal may provoke an additional response for this undesired modulation frequency that has to be avoided with
regard to the frequency dependence of the impedance
specific variables. The voltage amplitude is in the range
of some volts, the current amplitude is in the range of some
microamps to milliamps (mA to mA). The changes in the
impedance caused by volume changes can be very small,
<0.001%. This means that very small changes in the
measured current or voltage have to be processed. Hence,
the sensitivity and stability of the input amplifier must be
very high in order to detect such small changes in the
measured signal.
Independent from the measurement method, careful
consideration of measurement errors is necessary. A main
source of measurement errors may be parasitic components, such as stray capacitances between neighboring
wires leading to the sensing electrodes, or between wires
and their shielding, or stray capacitances between metallic
components of the measuring system and ground, which
become the more effective the higher the measuring
frequency.
The risk for undesired stimulation of the heart or peripheral nerves if such electrical voltages or currents are
applied for monitoring purposes, is negligible, both with
regard to the high frequency and the low current density.
Furthermore, heating and heat-induced secondary effects
can be neglected.
However, proper attention must be paid for the selection
of the equipment and its performance data for the intended
application. Furthermore, the employed devices must be
safe even in case of technical failure. Patient-near devices
are directly connected with the patient whereby the connecting impedance is rather low.
CHARACTERISTICS OF BIOIMPEDANCE
The microscopic electrical properties that describe the
interaction of an electromagnetic wave with biological
tissue are the complex conductivity s with the unit
V1m1, mhom1, Sm1, or 1(Vm)1
s v s 0 v js 00 v
and the complex dielectric permittivity e with the unit F/m
e v e0 v je00 v
v is radian frequency with the unit hertz. The electrical
properties are depending on the frequency with strong
dependence in the range of a dispersion.
The relation between these two quantities can be
described in accordance with Ref. 13 by
s v jv e v
104
(a)
(b)
Z Z0 e jw
or by its real part (or resistance) Re{Z} and its imaginary
part (or reactance) Im{Z}:
Z RefZg j ImfZg
Alternating current voltages and ac currents, too, can be
expressed as complex quantities, that is, by their magnitude and phase angle, or by their real and imaginary part
although this is rather unusual. The magnitude of the
impedance Z0 corresponds to the quotient of the magnitudes of voltage V0 and current I0, that is,
Z0 V0 =I0
Appropriate modeling of the electrical properties of biological tissue by discrete and lumped electrical components
renders possible the proper consideration of multilayer or
compartmentally composed tissues with different electrical
properties of each layer or compartment. Such tissues can
(c)
103
102
101
100
101
102
103
104
f [Hz] in logarithmic scaling
105
106
100
101
102
103
104
105
106
100
101
102
103
104
105
106
100
101
102
103
104
105
106
0
20
40
60
101
6000
4000
2000
0
101
0
1000
2000
3000
101
Imaginary Part Im {e } = e
The complexity of these quantities considers the fact that

in the alternating current (ac) range the biological tissue
cannot adequately be described by a simple resistance (or
its inverse conductance), but needs the extension to a
complex quantity, that is, impedance or admittance. Some
authors prefer the term Admittance Plethysmography
instead of Impedance Plethysmography (14,15). The
simplest adequate model for such an impedance is represented by a resistance and a reactance. The resistance
causes the loss in power, whereas the reactance causes
the delay (or 18 phase shift) between voltage and current.
The dominating reactance of bioimpedance in the frequency range of interest is capacitive and becomes more
relevant for higher frequencies. Only for very high frequencies that usually are not employed for impedance
plethysmography, can the reactance be composed by both
a capacitive and an inductive component.
Bioimpedance can be described like any technical impedance in different forms, for example, by its magnitude (or
modulus) Z0 and its phase angle (or argument) w, (i.e., the
delay between voltage and current):
Magnitude
r v r0 v j r00 v
Phase (degrees)
where e0 is the dielectric permittivity of free space with

e0 8.85
1012 Fm1, and er is the relative dielectric
permittivity (with er 1 for the free space and er 81 for
water in the low and medium frequency range).
Instead of the complex conductivity s, the inverse
complex resistivity r with the unit Vm can be used.
The resistivity is usually preferred in the context of impedance plethysmography:
Real Part Re {Z}
s 0 s 0 v e00 v
s 00 v e0 v e0 er v
be modeled as serial, parallel, or serialparallel equivalent

circuits in 2D presentation. More recently, the modeling
has been extended to 3D models using the FEM or comparable approaches.
The impedance parameters can be depicted in different
modes as a function of frequency (Fig. 4). The presentation
of the magnitude (usually in logarithmic scale with regard
to its wide range) and the phase angle against the frequency over several decades, and therefore in logarithmic
scale is known as the Bode plot. A similar presentation is
used for both the real and imaginary part versus the
frequency on the x axis. This mode of presentation is
sometimes called the spectrum (e.g., modulus spectrum
and phase angle spectrum). A different form of presentation is in a plane with the real part along the x axis and the
imaginary part along the y axis, both in linear scaling, with
the frequency as parameter. This mode of presentation is
frequently called the ColeCole-plot (but also the Nyquist
plot, locus plot, or Wessel graph). The same modes of
Imaginary Part Im {Z}
With the conduction current that is related with the basic

conductivity s0, that is, the current carried by the mobility
of ions in the extracellular space, and the polarization
current (sometimes called displacement current) that is
related with permittivity, the following equations are
obtained
125
3000
2000
1000
0
0
=0
=
2000
4000
Real Part Re {e } = e
Figure 4. Different modes for the presentation of impedance

quantities. (a) Shows the magnitude (in logarithmic scaling)
and the phase angle of impedance as functions of frequency in
logarithmic scaleing (Bode plot). (b) Shows the real and imaginary
part of impedance depicted versus the frequency in logarithmic
scaling. (c) Presents the ColeCole plot with the real part at the x
axis, the imaginary part at the y axis, and the frequency as
parameter along the curve. The results shown are not from
biological tissue, but calculated for the circuit of Fig. 1b with
Rs 500 V, Rp 5 kV, and Cp 500 nF.
126
presentation are possible for the complex quantities s, r,

and e.
Usually, impedance plethysmography is accomplished
employing only a single measuring frequency or a few
discrete measuring frequencies. However, impedance
spectroscopy with a multitude of measuring frequencies
is gaining interest, especially for the determination of
spatially distributed volumes. Typical examples are the
determination of body composition, tissue, or organ
vitality monitoring in combination with cellular edema
as a result of hypoxemia, and monitoring of infusion
therapy.
Certain forms of electrotherapy are also utilizing the
passive electrical properties of biological tissue. Methodology and technology for these forms of electrotherapy, however, are not discussed here.
MODEL-BASED RELATIONS FOR VOLUME
DETERMINATION
Valid relations between the monitored electrical quantities
and the searched volumetric parameters have to be used to
calculate a quantitative result that can be expressed in
units of volume (e.g., mL or cm3). Most of these relations
are based on models. Possibly the first model for the
interpretation of bioimpedance measurements has already
been developed for a suspension of cells in a fluid by Fricke
and Morse in 1925 (16).
The simplest model-based approach for establishing an
impedancevolume relationship is a cylindrical volume
conductor of radius r0, length L, and resistivity r*. It is
assumed that this volume conductor is surrounded by soft
material with significantly higher resistivity re* r*, so
that it must not be considered as a parallel circuit and its
actual radial extension has no impact. This volume conductor may be a blood vessel (e.g., an artery or vein)
surrounded by tissue that has higher resistivity than blood.
Furthermore, it is assumed that the inflow of the volume
DV into that cylinder expands the radius homogeneously
by Dr over the total length. It is also assumed that neither
the length L nor the resistivity r are affected by the
volume injected into the volume conductor. For didactic
simplicity, only the real part r0 of the complex resistivity r
is considered; that is, only the real part of the impedance is
taken into account. However, despite this simplification,
the variable is understood as impedance Z. This simplification is generally valid for frequencies much lower than the
b-dispersion, since for these frequencies the phase angle is
small (< 108). For higher frequencies, the calculation must
be performed with proper consideration of the complex
quantities.
With these assumptions the following relations for Z
and V are valid:
Z0 r0 L=pr02
V0 Lpr02
From these two equations the following equation can
easily be calculated
Z0 r0 L2 =V0
After the inflow of the volume DV and the increase of the

radius by Dr, the following relations are valid:
Z1 Z0 DZ r0 L=pr0 Dr2
V1 V0 DV Lpr0 Dr2
with Z1 < Z0 and V1 > V0 for Dr > 0.
With some simple mathematical operations it can be
shown that
r0 L2 Z r0 L2 DV Z Z DZ
If the product DV DZ Z is neglected as a product of small
quantities, the result becomes:
DV r0 L=Z2 DZ
This is the well-known Nyboer equation that relates the
volume change DV with the change in impedance DZ as a
consequence of the blood inflow into a peripheral artery,
for example, into the aorta or carotid artery with each
heart beat.
For proper application all included simplifications must
carefully be taken into account. Only mathematical simplifications, but no methodological constraints have been
mentioned here. Such a constraint may be that a part of the
injected volume is already flowing out of the measured
vessel segment before the inflow of the whole volume DV
into this vessel segment has been completed. This constraint must especially be considered if the segment is
short and the vessel wall rather stiff.
With regard to the surrounding tissue, a more realistic
model would be an arrangement of two concentric cylinders of length L with different conductivities. The inner
cylinder with radius r1 and resistivity r1 has the impedance Z1 r01 L=p r21 , whereas the outer cylinder with
radius r2 and resistivity r2 has the impedance Z2
r02 L=pr2 r1 2 . Electrically both cylinders are arranged
in parallel configuration. Hence, the total impedance is
obtained by Z0 Z1 Z2 =Z1 Z2 . The inner cylinder shall
be a vessel into which blood of volume DV is pumped,
causing a homogenous dilation of the vessel radius by Dr1
and a lowering of its impedance to Z1 Z1 DZ1
r01 L=pr1 Dr1 2 . Since Z2 shall not be affected, the total
impedance is Z0 Z1 Z2 =Z1 Z2 . The following steps
are similar to those leading to the Nyboer equation. Since
the resulting equation and its application to measurements
are more complicated, they will not be discussed here in
detail. Even this model is actually simplified, because in
reality the tissue around the vessel will neither be a cylinder
nor have a homogeneous resistivity. This last situation may
become relevant with a vein or a bone in the vicinity of the
artery.
With regard to the constraints of the Nyboer equation,
another approach has been used that finally leads to the
Kubicek equation (17). The model-based approach starts
again with the single-vessel model of length L. However, in
contrast to the Nyboer approach the assumption is not
made that the inflow of the volume DV into the considered
vessel segment is finished before the outflow starts. Here,
the basic assumption is that the inflow is constant during
the inflow time Tinf and that the outflow starts with some
delay, however, temporal overlap of outflow with inflow
must not be excluded. With this assumption, the change in

the intravascular volume and, hence, in the impedance, is
maximal when there is only inflow into and no outflow from
the segment. This maximal change of the impedance can be
expressed by its first time derivative [i.e., by (dZ/dt)max].
The total inflowing volume DV can then be taken into
account by multiplying (dZ/dt)max with the inflow time
Tinf. With regard to the aorta this inflow time is equivalent
with the ejection time of the left ventricle. In many cases
even the inflow time can additionally be obtained from the
impedance curve. This leads finally to the Kubicek equation:
DV r0 L=Z2 Tinf dZ=dtmax
Obviously, the only relevant difference in both approaches
is the Nyboer assumption that the total volume change DV
has already been injected into the measured vessel segment before the outflow starts against the Kubicek
assumption that this volume DV is entering the measured
vessel segment with constant rate during the whole inflow
period. The Kubicek equation is more realistic for a short
vessel segment with a rather stiff wall. For such vessels,
the Nyboer equation leads to an underestimation of the
real volume change. In contrast, if the inflow is decreasing
at the end of the inflow period, for example, at the end of the
ventricular ejection period, the Kubicek equation yields an
overestimation of the volume change.
All other model-based assumptions are identical or
comparable. Both approaches consider only a single vessel
with homogeneous dilation over the total length within the
measured tissue and neglect the surrounding tissue and its
composition with regard to nonhomogeneous resistivity.
Blood resistivity is taken as constant although there is
some evidence that it depends on the flow velocity.
Although the Kubicek equation has primarily been
proposed for the monitoring of cardiac output, both equations have also been applied to the monitoring of pulsatile
peripheral blood flow. Both models, however, do not consider that in the peripheral circulation a basic or nonpulsatile blood flow may exist as well.
Different modifications have been suggested in order to
overcome relevant drawbacks. Most of these modifications
are optimized with regard to the monitored quantity, geometric constraints, modes of application, or positioning and
shape of electrodes. They will not be discussed here.
No valid impedancevolume models have been proposed
for the quantitative monitoring of ventilation by the application of impedance plethysmography. Statistical models
are used for the impedancevolume relationship regarding
body composition. Some first approaches have been suggested for the volume changes due to fluid shifts.
INSTRUMENTATION AND APPLICATIONS
The typical basic equipment for impedance plethysmography consists of the signal generator, either a constant
voltage generator or a constant current generator; the
frequency-selective measuring device, either for current
or voltage, in combination with AD conversion. The equipment may be supplied with more than one signal channel
127
for certain applications, for example, with two channels for

simultaneous and symmetric monitoring at both extremities or one channel for each frequency in multifrequency
measurements; the signal processor, for example, for processing the impedance quantities; the processing unit for
calculating the volumetric quantities; the monitor and/or
data recorder; multiple electrodes and shielded leads; specific auxiliary equipment, for example, venous occlusion
machine with cuff and pump.
Devices for impedance plethysmography are small,
light, usually portable, and battery powered. The devices
for patient-near application are much cheaper than competitive equipment based on nuclear magnetic resonance
(NMR), X ray, or US. Also, the running costs are much
lower than for the competitive technologies, usually these
costs are mainly required by the single-use electrodes.
Peripheral Hemodynamics
The objective is the detection of deficiencies either in the
arterial or venous peripheral circulation. The application
of impedance plethysmography to peripheral vascular studies has already been in the interest of Nyboer in 1950 (18).
In the peripheral circulation, the most interesting quantity is arterial blood flow or perfusion. Impedance measurement is performed either in the bipolar or, more frequently,
in the tetrapolar configuration. The tetrapolar configuration requires a longer segment for measurement in order to
place the sensing electrodes in proper distance from the
feeding electrodes with the nonhomogenous current field in
their vicinity. Electrodes are either of the circular or disk or
the band type. Disk electrodes can be placed directly above
the monitored vessel and therefore provide high sensitivity, but the magnitude and the reproducibility of the measured signal in repeated measurements are strongly
dependent on exact electrode placement (1921). Band
electrodes are preferred for the measurements at extremities because they can be placed around the extremities. In
this case, the measured object is the whole segment
between the sensing electrodes including the extravascular
tissue and may include more than only one vessel. Flow can
be estimated by application of the Nyboer, the Kubicek or
any modified impedancevolume equation. Competitive
methods are utilizing ultrasound Doppler, contrast X-ray
angiography, or NMR.
Some diagnostic information about the stiffness of the
arterial vessel wall can be obtained by the impedance
method from the measurement of the pulse wave propagation velocity, usually executed at two different sites of the
same arterial pathway. The pulse that is actually recorded
with the impedance method is the intravascular volume
pulse, that is, the dilation of the vessel with each heart beat
(22). Simple formalistic models are used to relate the pulse
wave propagation velocity with the stiffness of the vessel
wall. If the intravascular blood pressure is also monitored,
then it is possible to calculate the stiffness or its inverse,
the compliance as ratio of the volume change and pressure
change DV/Dp, directly.
Another problem is the diagnosis of proximal or deep
venous thrombosis and of other obstacles for the venous
return flow to the heart from the extremities (23). One
128
approach is actually a modification of the venous occlusion

plethysmography that has already been introduced in 1905
by Brodie and Russel (24). A cuff is placed around the
extremity and connected with a pump. The cuff pressure is
enhanced abruptly so that it occludes the vein and stops
venous outflow without affecting the arterial inflow. The
volume increase following venous occlusion allows estimating the arterial inflow. When the occlusion is stopped after
20 s, the venous outflow starts again and thereby leads
to a reduction in volume. The slope or the time constant of
this postocclusion emptying process are used to assess the
outflow resistance, for example, the hydrodynamically
obstructive impact of deep venous thrombosis, or the
venous wall tension. However, other pathological effects
must be carefully considered, for example, increased central venous pressure. For this reason, the recording is
frequently and simultaneously performed on both extremities, so that the results can be compared with each other.
The measurement is usually executed with band electrodes. Competitive methods are ultrasound Doppler, contrast X-ray venography, or NMR.
A similar impedance-based approach is employed to test
the performance of the drainage system in extremities.
Changes in the hydrostatic pressure are used to shift
volume between the trunk and an extremity, for example
first by bringing down an arm before raising it above the
head. The affected volume shifts can be recorded and
render possible the assessment of the performance of the
draining system. This approach is frequently used to study
fluid shifts caused by tilting experiments, during microgravity experiments, or after long periods of bedrest.
Brain Perfusion and Edema
The most important objectives are monitoring of cerebral
bloodflow and the detection of cerebral edema. First publications about rheoencephalography are dating back to
1965 (25,26).
The volume of the brain with its enclosed fluid spaces,
for example, the intravascular volume and the cerebrospinal fluid volume, is kept constant by its encapsulation in
the bony skull. The expansion of the volume of one compartment, for example, increase of the intravascular
volume by augmented arterial blood pressure, the spacedemanding growth of a brain tumor or intracerebral bleeding, can only be compensated by the diminution of the
volume of other compartments. If the space-demanding
process is of nonvascular origin, the most affected compartment will be the intravascular volume. Due to the compression of blood vessels, the cerebral bloodflow and thus
metabolism will be reduced.
Impedance measurements aiming for the brain as organ
are difficult because the encapsulating bony skull has a
very high resistivity as compared with the soft extracranial
tissue of the face and the scalp. If the tetrapolar mode is
used, more than two sensing electrodes may be applied.
Different electrode arrangements have been described to
force the current pathways through the skull into the
brain, but also the application of the feeding electrodes
to the closed eyelids. However, the measurement of the
transcephalic impedance has not become a routinely
applied clinical method with the exception of neonates in

which the thickness of the bony skull is very small. Competitive methods based on NMR, X ray, US, and even
photoplethysmography have gained more attention in
the recent past. Some expectations are related to the
development of contactless applications, especially for
the continuous monitoring of edema (27). This might allow
control treatment by hyperosmolaric infusion therapy.
Ventilation and Lung Performance
Impedance pneumography were among the first applications of impedance plethysmography and had already been
described by Geddes et al. in 1962 (28,29).
The objective of impedance pneumography is to record
the tidal volume under resting conditions or during exercise. Additionally, the breathing rate can be obtained by
the impedance method. The principle is based on the
measurement of the transthoracic impedance that
increases during inspiration as a consequence of increasing
alveolar air filling, and decreases during expiration (30).
The conductivity of lung tissue at the end of a normal
expiration is 0.10 V1m1 as compared with
0.05 V1m1 at the end of normal inspiration. The application of impedance pneumography is very simple and also
applicable for critically ill patients, since it allows continuous recording without requiring a breathing tube. However, the quantitative determination of the tidal volume is
difficult and needs calibration by another spirometric
method. No realistic model-based interpretation of quantitative assessment is available until now. Some expectations are related with multifrequency measurement (31).
The impedance measurement can be performed with the
bi- or tetrapolar configuration. It is usually performed
separately for each side of the lung in order to detect
differences. Even with more electrodes, however, the spatial resolution is too poor to allow detection of regional
inhomogeneities of alveolar air filling. For that reason, this
field is gaining growing interest for the application of EIT
(4). Also, EIT has limited spatial resolution, as compared
with other CT-based imaging procedures. Despite this
drawback, it has some potential for the detection of regional inhomogeneities in ventilation. Such an approach
would be of high relevance for diagnostic purposes and
for the efficiency control of artificial ventilation. Serious
drawbacks for EIT, however, are the costs of the equipment
and the necessity to attach up to 64 or 128 electrodes
around the thorax.
Since pulmonary edema is usually a general and not a
localized phenomenon, its monitoring might be possible by
utilizing the transthoracic impedance measurement. Measurement of extravascular lung water is investigated as a
methodological approach to guide the fluid management of
patients with noncardiogenic pulmonary edema (32). The
conductivity of lung edema fluid is 1 V1m1, and therefore is distinctly different from alveolar tissue filled with
more or less air.
The impedance coefficients of tumor tissue are different
from normal lung tissue. Hence, cancer detection might be
possible by bioimpedance measurement. But with regard to
its poor spatial resolution, the bi- or tetrapolar transthor-
acic impedance measurement is not qualified, but EIT

might become useful for some special applications. Other
imaging procedures were superior until now, primarily due
to the higher spatial resolution as compared with EIT.
Much work has been devoted to comparing impedance
pneumography with inductive pneumography. There is
some evidence that inductive pneumography is superior
concerning ventilation monitoring in newborn infants,
especially for risk surveillance with regard to SIDS. An
interesting approach tries to utilize inductive pneumography in combination with the evaluation of the signal morphology for the monitoring of airway obstruction (33).
129
Head
Neck
Torso
Upper arm
Intercompartmental Fluid Shifts

Intercompartmental fluid shifts occur during dialysis, for
example, hemodialysis, but also during infusion therapy
and emergence of edema. The assessment of fluid shifts,
which are actually changes of volumes, by the measurement of electric variables has been an outstanding objective very early in the scientific research and medical
utilization of bioimpedance and dates back to 1951 (34).
Hemodialysis is a therapeutic procedure that is employed in patients with renal insufficiency. The therapeutic
objective is to remove water, electrolytes, urea, and other
water-soluble substances in combination with the reestablishment of a normal acidbase status. In a simplified
model, the water is first removed from the intravascular
volume, (i.e., the blood plasma). This means that the
hematocrit, and thereby the viscosity of blood, is increased
cousing the work load for the heart is enhanced. Also, the
osmotic pressure of the blood is raised, whereas the hydrostatic blood pressure is lowered. Both effects contribute to
the refilling of the intravascular space by a fluid shift from
the interstitial space (i.e., the extravascular extracellular
space). This fluid shift finally causes another fluid shift
from the intracellular space into the interstitial space. The
dynamics of these fluid shifts is primarily controlled by the
hydrostatic and osmotic pressure differences between the
different spaces, but also by the substance-specific permeability of the different barriers between the spaces including the dialysis membrane. If removal of water is too fast or
changes of ions like sodium, potassium, and calcium are too
large, the hemodynamics of the patient or the excitability of
tissues like the heart or central nervous system may
become disturbed. Impedance plethysmographic methods
have some potential for the control of those fluid shifts, that
is, may help to avoid critical disequilibrium syndromes like
hypotension, headache, and vomiting. The best results of
the plethysmographic measurement are achieved if segmental measurement is performed at the extremities
instead of whole body measurements (5). Figure 5 shows
a schematic presentation of such segmented body. The
forearm accounts only for 1% of body weight, but contributes 25% to whole body impedance.
Infusion therapy aims mainly to filling the intravascular volume by utilizing venous access. However, depending
on the control variables (e.g., hydrostatic pressure, osmotic
pressure, and permeability of the barriers), intercompartmental fluid shift cannot be avoided. Consequently, part of
the infused volume will not remain in the intravascular
Lower arm
Upper leg
(thigh)
Lower leg
(calf)
Figure 5. Schematic presentation of the whole body subdivided

into 11 segments that are modeled by RC-networks.
system and help to stabilize blood pressure, but escape to

the extravascular space.
A special objective of intravenous infusion therapy with
hyperosmolaric fluid is the removal of fluid from the extravascular space in order to avoid the emergence of edemas.
In the brain, such infusion may help to lower the intracranial pressure that is caused by edema.
Body Composition
Most frequently this method is called bioelectric impedance
analysis (BIA). The interest in this methodological
approach has increased with the increased interest in
healthy life style in industrialized populations since the
first reports at the end of the 1980s (35,36).
The body consists of different tissues, for example, muscle, fat, bones, skin, the nervous system, and connective
tissue. All these tissues contain intra- and extravascular
spaces. The extravascular space can be subdivided into the
extracellular or interstitial space and the intracellular
space. Chemically, the body consists of water, proteins,
carbohydrates, lipids, ions, rare elements, and some other
substances. The main objective of body composition analysis
130
is the assessment of total body water, of lean or fat free

tissue mass, and of fat tissue. Those measurements have
clinical relevance, but they are also gaining growing attention in the field of sports, physical fitness, wellness, nutrition, and for life science in aerospace research.
The methodological approaches are different, for example bipolar or tetrapolar mode, single-or multiplefrequency measurement, whole body or segmental measurement. In single-frequency measurements, a sinussoidal current with usually < 1 mA (range 0.0210 mA) and a
frequency with typically 50 kHz (range 20100 kHz) is
employed. The basic assumption is that resistivity of fat
tissue is higher than that of the so-called lean tissue that
has a higher content of water. In the nonclinical field, the
determination of body composition is frequently combined
with the measurement of body weight using a scale with
two integrated electrodes in the foot platform. In that case,
however, only the two legs and the lower part of the trunk
will be included into the estimation. In more advanced
devices, additional electrodes are available in the form of
hand grips. Detailed segmental measurements are more
valid since the trunk of the body may contribute 70% to
the total body weight and up to 90% to the total body fat,
but only 5% to the measured whole body impedance. More
sophisticated equipment utilize multifrequency measurement (11).
The applied statistically based predictive equations are
of the linear regression type and consider individual parameters like sex, age, height, and weight. They are primarily used to assess the lean muscle mass, the body fat, the fat
free mass, the water content, and the body mass index. For
physically based segmental calculations, the extremities,
the body trunk, and even the head are modeled by cyclindric shape with uniform cross-section over the total length.
Competitive methods include antropometric measures
like the girth, simple skin-fold measurements by mechanical calipers, but also highly advanced methods, such as
NMR, X ray (dual energy X-ray absorptiometry, or nuclear
imaging), and for certain purposes even the hydrostatic
weighing in a water tank that is assumed to be the most
accurate method.
Laboratory Applications
Blood Cell Counting. The employed methodological
principle is frequently called the Coulter principle (37).
Blood cells (e.g., erythrocytes, leucocytes) are passing
through a capillary (diameter < 100 mm) filled with blood
plasma. For frequencies below the b-dispersion the impedance of the cell is higher than that of the surrounding
plasma. Consequently, the passage of each cell affects the
recorded impedance. Those impedance changes are used
for cell counting. In sophisticated devices the impedance
changes are quantitatively measured and allow cell volume
estimation, which also renders possible the determination
of the cell volume distribution function (frequently called
the Price Jones distribution function with the cell diameter as variable). Since the number of leucocytes is very
small compared with that of the erythrocytes (usually
< 0.1%), leucocytes do not really disturb the counting of
erythrocytes. In contrast, however, the erythrocytes must
be destroyed before the leucocytes can be counted. This is

usually achieved by chemical hemolysis of the erythrocytes. Furthermore, chemical substances are utilized to
render possible the differentiation between the different
populations of leucocytes (e.g., granulocytes, monocytes,
lymphocytes).
Hematocrit. The objective is the estimation of the
intracellular volume of blood. The hematocrit is defined
as the ratio of the volume of blood cells to total blood
volume, although frequently the hematocrit is taken as a
measure for the ratio of only the volume of erythrocytes to
total blood volume. However, since the partial volume of all
other blood cells (i.e., leucocytes, platelets) is very small
compared with that of the erythrocytes, the results are not
distinctly different. Determination of the ratio between the
extracellular volume and the total blood volume is possible
by application of at least one measuring frequency below
and another one above the b-disperion. Problems that need
further consideration as possible sources of error are the
electrolytic conductivity of blood plasma, which is dependent on the actual concentration of ions, and the sedimentation of the cells as a consequence of their higher specific
weight that may take place during the measuring period.
The measurement requires the withdrawal of a sample of
blood from a vein or another blood vessel. Measurement of
the hematocrit can be achieved either with the impedance
or the dielectric technique (38).
Several modifications of the impedance method for the
noninvasive determination of the hematocrit have been
proposed. In one approach, impedance measurement is
performed at the finger by application of two different
frequencies (i.e., 100 kHz and 10 MHz). The hematocrit
is determined by an algorithm that uses both the pulsatile
and the baseline component of both measuring frequencies
(39). In another remarkable methodological approach, the
patient puts a finger in a temperature-stabilized bath. A
fluid with gradually increasing ionic concentration is
pumped through the bath chamber, thus leading to a
decrease in the impedance of the bath fluid. The pulsatile
volume changes of the finger are recorded by impedance
measurement in the bath chamber. These pulsatile impedance fluctuations disappear only if the impedance of the
bath fluid is identical with that of the blood in the finger.
The conclusion is made from the actual impedance of the
bath fluid on the hematocrit (40).
Others
Cell Imaging. Nanotechnology-based on bioimpedance
sensing, chip devices have been described that allows us
to rapidly detect and image cells with a specific phenotype in a heterogeneous population of cells (41). This might
be useful for screening purposes, recognition of cell irregularities, and detection of risk patients like human
immunodeficiency virus (HIV)-infected individuals. Cell
identification is made possible by administration of marker
substances. A promising measurement procedure is electrochemical cyclic voltammetry. When a sinusoidal voltage
with constant amplitude and a variable frequency in the
range of some kilohertz is applied, the impedance is plotted
in the spectrographic mode. Among other effects, volume

changes might be the dominating measured effect if the
marker binds with a receptor in the cell membrane, and
herewith affects membrane properties like its permeability
for water or ions or the transmembraneous active ion
transport mechanisms.
Inductive Plethysmography. Inductive plethysmography is employed for respiratory monitoring, so-called
respiratory inductance plethysmography (RIP). It is based
on the measurement of the thoracic crosssection that is
enclosed by coils and includes both the rib cage and abdominal compartments (4245). In the medium frequency
range (30500 kHz), changes in the volume are monitored
by the influenced inductance. In the higher frequency
range (100 MHz), the inductively provoked signals (i.e.,
eddy currents depending on the alveolar air filling) are
recorded by appropriately arranged coils.
Magnetic Susceptibility Plethysmography. Magnetic susceptibility plethysmography is a contactless method. It is
based on the application of a strong magnetic field and
monitors the variation of the magnetic flux. The measurement is accomplished with superconduting quantum interference device (SQUID) magnetometers. This approach
may primarily be utilized for the assessment of blood
volume changes in the thorax, but until now it is not
employed for clinical routine (46).
SUMMARY
In comparison with biomedical engineering as a recognized
discipline, the research activities in the field of bioimpedance
are much older. It can be assumed that Nikola Tesla, a former
student of physics in Graz (Austria) and the inventor of the ac
technology, already knew about the passive electrical properties of biological tissues when he demonstrated his famous
and public performances with the administration of high
voltage pulses I demonstrated that powerful electrical
discharges of several hundred thousand volts which at that
time were considered absolutely deadly, could be passed
through the body without inconvenience or hurtful consequences in the 1880s. This knowledge was utilized by
dArsonval since 1892 for therapeutic purposes, mainly aiming for heat induction in certain parts of the body. In 1913,
Rudolf Hoerber, at that time a physiologist at the University
of Kiel (Germany), measured the electrical conductance of
frog muscle at 7 MHz and found that at this frequency the
membrane resistance is short circuited.
Since its beginning, bioimpedance remained to be a
challenge to physicists, medical doctors, and of course
engineers. The most relevant basic research was performed
in the second half of the twentieth century. The progress
that has been reached has been and is still utilized both for
diagnostic and therapeutic purposes in medicine. Impedance plethysmography is one of the different fields of
bioimpedance application. If impedance plethysmography
is correctly understood, it does not only mean the determination of a solid volume with well-defined boundaries, but
also the volumetric determination of one component contained in a mixture of different components.
131
Progress in technology has rendered possible applications that are of great interest for medicine. The most
relevant progress is in the field of signal acquisition,
including advanced electrode technology, signal processing, and model-based signal interpretation. Not all
attempts to utilize the passive electrical properties of
biological tissue for diagnostic purposes have been successful. In many cases, other technologies have been shown to
be superior. But there is no doubt that the whole potential
of impedance plethysmography has not been exhausted.
New challenges in the medical field are cellular imaging
and possibly even molecular imaging. In all applications,
however, impedance plethysmography will have to prove
its validity and efficiency.
BIBLIOGRAPHY
1. Nyboer J, Bango S, Nims LF. The Impedance Plethysmograph
and Electrical Volume Recorder. CAM Report OSPR 1943;
149.
2. Schwan HP. Electrical properties of body tissues and impedance plethysmography. IRE Trans Biomed Electron 1955;
3:3246.
3. Stewart GN. Researches on the circulation time in organs and
on the influence which affect it. J Physiol 1894;15:189.
4. Li J. Multifrequente Impedanztomographie zur Darstellung der
elektrischen Impedanzverteilung im menschlichen Thorax.
PhD thesis, University of Stuttgart, 2000. Available at http://
elib.uni-stuttgart.de/opus/volltexte/2000/736/pdf/li_diss.pdf.
5. Kanai H, Haeno M, Sakamoto K. Electrical measurement of
fluid distribution in legs and arms. Med Prog Technol
1987;12:159170.
6. Osten H. Impedanz-Plethysmographie im Orthostasetest.
Mu nchn Med Wochenschr 1977;119:897900.
7. Gersing E. Measurement of electrical impedance in organs.
Biomed Techn 1991;36:611.
8. Dzwonczyk R, et al. Myocardial electrical impedance responds to ischemia in humans. IEEE Trans Biomed Eng
2004;BME-51:22062209.
9. Durney CH, Massoudi H, Iskander MF. Radiofrequency
Radiation Dosimetry. 4th ed. USAFSAM-TR-8573; 1985.
10. Gabriel S. 1997. Appendix B: Part 1: Literature Survey. Available at http://niremf.ifac.cnr.it/docs/DIELECTRIC/AppendixB1.
html.
11. Segal KR. Estimation of extracellular and total body water
by multiple-frequency bioelectrical impedance measurement.
Am J Clin Nutr 1991;V54-1:2629.
12. Neuman MR. Biopotential electrodes. In: Webster JG, editor.
Medical InstrumentationApplication and Design. 3rd ed.
New York: Wiley; 1998.
13. Rigaud B, Morucci J-P, Chauveau N. Bioimpedance measurement. In: Morucci J-P, et al. edition. Bioelectrical Impedance
Techniques in Medicine. Crit. Rev. Biomed. Eng. 1996;
24: 257351.
14. Yamakoshi KI, Shimazu H, Togawa T, Ito H. Admittance
plethysmography for accurate measurement of human limb
blood flow. Am J Physiol 1978;235:H821H829.
15. Shimazu H, et al. Evaluation of the parallel conductor theory
for measuring human limb blood flow by electrical admittance
plethysmography. IEEE Trans Biomed Eng 1982;BME-29:
17.
16. Fricke H, Morse S. The electrical resistance of blood between 800 and 4.5 million cyclces. J Gen Physiol 1925;9:
153157.
132
17. Kubicek WG, et al. Development and evaluation of an impedance cardiac output system. Aerospace Med 1966;37:1208
1212.
18. Nyboer J. Electrical impedance plethysmography: a physical
and physiological aproach to peripheral vascular studies.
berg PA. Impedance plethysmo19. Yamamoto Y, Yamamoto T, O
graphy in human limbs. Part 1: On electrodes and electrode
geometry. Med Biol Eng Comput 1991;29:419424.
berg PA. Impedance plethysmo20. Yamamoto Y, Yamamoto T, O
graphy in human limbs. Part 2: Influence of limb crosssectional areas. Med Biol Eng Comput 1992;30:518524.
21. Lozano A, Rosell J, Pallas-Areny R. Errors in prolonged electrical impedance measurement due to electrode repositioning
and postural changes. Physiol Meas 1995;16:121130.
22. Risacher F, et al. Impedance plethysmography for the evaluation of pulse wave velocity in limbs. Med Biol Eng Comput
1992;31:318322.
23. Hull R, et al. Impedance plethysmography: The relationship
between venous filling and sensitivity and specificity for proximal vein thrombosis. Circulation 1978;58:898902.
24. Brodie TG, Russel AE. On the determination of the rate of blood
flow through an organ. J Physiol 1905;33:XLVIIXLVIII.
25. Seipel JH, Ziemnowicz SAR, ODoherty DS. Cranial impedance plethysmographyrheoencephalography as a method
for detection of cerebrovascular disease. In: Simonson E,
McGavack TH, editors. Cerebral Ischemia. Springfield, IL:
Charles C Thomas; 1965; p 162179.
26. Hadjiev D. A new method for quantitative evaluation of cerebral blood flow by rheoencephalography. Brain Res
1968;8:213215.
27. Netz J, Forner E, Haagemann S. Contactless impedance measurement by magnetic inductiona possible method for investigation of brain impedance. Physiol Meas 1993;14:463471.
28. Geddes LA, Hoff HE, Hickman DM, Morre AG. The impedance
pneumograph. Aerospace Med 1962;33:2833.
29. Baker LE, Geddes LA, Hoff HE. Quantitative evaluation of
impedance spirometry in man. Am J Med Elect 1965;4:7377.
30. Nopp P, et al. Dielectric properties of lung tissue as a function
of air content. Phys Med Biol 1993;38:699716.
31. Brown BH, et al. Multifrequency imaging and modelling of
respiratory related electrical impedance changes. Physiol
Meas 1994;15(Suppl. A):112.
32. Nierman DM, et al. Transthoracic bioimpedance can measure
extravascular water in acute lung injury. J Surg Res
1996;65:101108.
33. Brack Th et al. Continuous and cooperation-independent
monitoring of airway obstruction by a portable inductive
plethysmograph. AJRCCM 2004;169:1.
34. Lo fgren B. The electrical impedance of a complex tissue and its
relation to changes in volume and fluid distribution. Acta
Physiol Scand 1951;23(Suppl. 81):151.
35. Schloerb PR, et al. Bioimpedance as a measure of total body
water and body cell in surgical nutrition. Eur Surg Res
1986;18:1.
36. van Loan MD, et al. Association of bioelectric resistive impedance with fat-free mass and total body water estimates of
body composition. Amer J Human Biol 1990;2:219226.
37. Coulter WH. High speed automatic blood cell counter and cell
size analyzer. Proc Nat Electron Conf 1956;12:1034.
38. Treo EF, et al. Comparative analysis of hematocrit measurements by dielectric and impedance techniques. IEEE Trans
Biomed Eng 2005;MBE-52:549552.
39. http://www.patentalert.com/docs/001/z00120411.shtml.
40. Yamakoshi KI, et al. Noninvasive measurement of hematocrit
by electrical admittance plethysmography technique. IEEE
41. Mishra NN, et al. Bio-impedance sensing device (BISD) for

detection of human CD4 cells. Nanotech 2004, vol. 1, Proc
2004 NSTI Nanotechnology Conf 2004, p 228231.
42. Brouillette RT, Morrow AS, Weese-Mayer DE, Hunt CE.
Comparison of respiratory inductive plethysmography and
thoracic impedance for apnea monitoring. J Ped 1987;111:
377383.
43. Valta P, et al. Evaluation of respiratory inductive plethysmography in the measurement of breathing pattern and PEEPinduced changes in lung volume. Chest 1992;102:234238.
44. Cohen KP, et al. Design of an inductive plethysmograph for
ventilation measurement. Physiol Meas 1994;15:217229.
45. Stro mberg TLT. Respiratory Inductive Plethysmography.
Linko ping Studies in Science and Technologies Dissertations
No. 417, 1996.
46. Malmivuo J, Plonsey R. Impedance plethysmography. In:
Malmivuo J, Plonsey R, editors, Bioelectromagnetism. New
York: Oxford University Press; 1995. Chapt. 25.
Further Reading
Foster KR, Schwan HP. Dielectric properties of tissue and biological materials. A critical review. Crit Rev Biomed Eng
1989;17:25102.
Kaindl F, Polzer K, Schuhfried F. (1958, 1966, 1979): Rheographie.
Darmstadt, Dr: Dietrich Steinkopff Verlag; 1958 (1st ed), 1966
(2nd ed), 1979 (3rd ed).
Morucci J-P, et al. Bioelectrical impedance techniques in medicine.
Crit Rev Biomed Eng 1996;24:223681.
Pethig R. Dielectric and Electronic Properties of Biological Materials. Chichester: Wiley; 1979.
Schwan HP. Determination of biological impedances. In: Nastuk
WL, editor. Physical Techniques in Biological Research. Vol. VI
(ptB) New York: Academic Press; 1963; p 323407.
Schwan HP. Biomedical engineering: A 20th century interscience.
Its early history and future promise. Med Biol Eng Comput
1999;37(Suppl. 2):313.
Stuchly MA, Stuchly SS. Dielectric properties of biological
substancestabulated. J Microw Power 1980;15:1926.
Webster JG, editor. Medical InstrumentationApplication and
Design. 3rd ed. New York: Wiley; 1998.
Gabriel C, Gabriel S. Compilation of the Dielectric Properties of
Body Tissues at RF and Microwave Frequencies, 1996. Available
at http://www.brooks.af.mil/AFRL/HED/hedr/reports/dielectric/
Report/Report.html.
See also BIOIMPEDANCE
IN CARDIOVASCULAR MEDICINE; PERIPHERAL
VASCULAR NONINVASIVE MEASUREMENTS.
BIRGITTE FREIESLEBEN DE BLASIO
JOACHIM WEGENER
University of Oslo
Oslo, Norway
INTRODUCTION
Impedance spectroscopy (IS), also referred to as electrochemical impedance spectroscopy (EIS), is a versatile
approach to investigate and characterize dielectric and
conducting properties of materials or composite samples
(1). The technique is based on measuring the impedance
(i.e., the opposition to current flow) of a system that is being

excised with weak alternating current or voltage. The
impedance spectrum is obtained by scanning the sample
impedance over a broad range of frequencies, typically
covering several decades.
In the 1920, researchers began to investigate the impedance of tissues and biological fluids, and it was early
known that different tissues exhibit distinct dielectric
properties, and that the impedance undergoes changes
during pathological conditions or after excision (2,3).
The advantage of IS is that it makes use of weak
amplitude current or voltage that ensures damage-free
examination and a minimum disturbance of the tissue.
In addition, it allows both stationary and dynamic electrical properties of internal interfaces to be determined,
without adversely affecting the biological system. The
noninvasive nature of the method combined with its high
information potential makes it a valuable tool for biomedical research and many medical applications are currently
under investigation and development; this will be reviewed
at the end of the article.
This article starts with providing a general introduction
to the theoretical background of IS and the methodology
connected to impedance measurement. Then, the focus will
be on applications of IS, particularly devises for in vitro
monitoring of cultured cell systems that have attracted
widespread interest due to demand for noninvasive, marker-free, and cost-effective methods.
THEORY
Impedance, Z, is a complex-valued vector that describes
the ability of a conducting medium to resist flow of alternating current (ac). In a typical IS experiment (Fig. 1), a
sinusoidal current I(t) signal with angular frequency v
(v 2pf ) is passed through the sample and the
resulting steady-state voltage U(t) from the excitation is
133
measured. According to the ac equivalent of Ohms law, the

impedance is given by the ratio of these two quantities
Z
Ut
It
(1)
The impedance measurement is conducted with use of

weak excitation signals, and in this case the voltage
response will be sinusoid at the same frequency v as the
applied current signal, but shifted in phase w. Introducing
complex notation, Eq. 1 translates into
Z
U0
expiw jZjexpiw
I0
(2)
with U0 and I0 being the amplitudes

of the voltage and
p
current, respectively, and i 1 being the imaginary
unit. Thus, at each frequency of interest the impedance
is described by two quantities: (1) its magnitude jZj, which
is the ratio of the amplitudes of U(t) and I(t); and (2) the
phase angle w between them. The impedance is measured
over a range of frequencies between hertz and gigahertz,
dependent on the type of sample and the problem to study.
The measured impedance can be divided into its real
and imaginary components, that is, the impedance contribution arising from current in-phase with the voltage and
908 out-of-phase with the voltage
R ReZ jZjcosw;
X ImZ jZjsinw
(3)
The real part is called resistance, R, and the imaginary

part is termed reactance, X. The reactive impedance is
caused by presence of storage elements for electrical
charges (e.g., capacitors in electrical circuit).
In some cases it is convenient to use the inverse quantities, which are termed admittance Y 1/Z, conductance
G Re(Y), and susceptance B Im(Y), respectively. In
the linear regime (i.e., when the measured signal is proportional to the amplitude of the excitation signal), these
two representation are interchangeable and contain the
same information. Thus, IS is also referred to as admittance spectroscopy.
INSTRUMENTATION
Figure 1. Schematic of a two-electrode setup to measure the

frequency-dependent impedance of a sample that is sandwiched
between two parallel plate electrodes.
The basic devices for conducting impedance measurements

consist of a sinusoid signal generator, electrodes, and a
phase-sensitive amplifier to record the voltage or current.
Commonly, a four-electrode configuration is used, with two
current injecting electrodes and two voltage recording
electrodes to eliminate the electrodeelectrolyte interface
impedance. As discussed below, some applications of IS
make use of two-electrode arrangements in which the same
electrodes are used to inject current and measure the
voltage.
Since the impedance is measured by a steady-state
voltage during current injection, some time is needed when
changing the frequency before a new measurement can be
performed. Therefore, it is very time consuming if each
frequency has to be applied sequentially. Instead, it is
common to use swept sine wave generators, or spectrum
analyzers with transfer function capabilities and a
white noise source. The white noise signal consists of the
134
superposition of sine waves for each generated frequency,

and the system is exposed to all frequencies at the same
time. Fourier analysis is then used to extract the real and
imaginary parts of the impedance.
The electrodes used for impedance experiments are
made from biocompatible materials, such as noble metals,
which in general is found not to have deleterious effect on
biological tissue function. Electrode design is an important
and complicated issue, which depends on several factors,
including the spatial resolution required, the tissue depth,
and so on. It falls beyond the scope of this article to go
further into details. The interested readers are referred to
the book by Grimnes and Martinsen (4) for a general
discussion.
Common error sources in the measurements include
impedance drift (e.g., caused by adsorption of particles
on the electrodes or temperature variations). Ideally, the
system being measured should be in steady-state throughout the time required to perform the measurement, but in
practice this can be difficult to achieve. Another typical
error source is caused by pick-up of electrical noise from the
electronic equipment, and special attention must be paid to
reduce the size of electric parasitics arising, for example,
from cables and switches.
DATA PRESENTATION AND ANALYSIS

The most common way to analyze the experimental data is
by fitting an equivalent circuit model to the impedance
spectrum. The model is made by a collection of electrical
elements (resistors, capacitors) that represents the electrical composition in the system under study.
As a first step, it is useful to present the measured data
by plotting log jZj and w versus log f in a so-called Bodediagram (Fig. 2a), and by plotting ImjZj versus RejZj
named a Nyquist diagram or an impedance locus (Fig.
2b). The examples provided in Fig. 2 are made for an
electrical circuit (insert Fig. 2b). While the first way of
presenting the data shows the frequency-dependence
explicitly, the phase angle w is displayed in the latter.
The impedance spectrum gives many insights to the
electrical properties of the system, and with experience it is
possible to make a qualified guess of a proper model based
on the features in the diagrams (cf. Fig. 4). Similar to other
spectroscopic approaches like infrared (IR) or ultraviolet
(UV)/visible (vis), the individual components tend to show
up in certain parts of the impedance spectrum. Thus,
variations in the values of individual components alter
the spectrum in confined frequency windows (Fig. 3).
For a given model, the total impedance (transfer function) is calculated from the individual components with use
of Ohms and Kirchhoffs laws. The best estimates for the
parameters, that is, the unknown values of the resistors
and capacitors in the circuit, are then computed with use of
least-square algorithms. If the frequency response of the
chosen model fits the data well, the parameter values are
used to characterize the electrical properties of the system.
In order to fit accurately the equivalent circuit model
impedance to the impedance of biomaterials, it is often
necessary to include nonideal circuit elements, that is,
Figure 2. Different representations of impedance spectra. (a and

b) visualize the frequency-dependent complex impedance of the
electrical circuit shown in (b) with the network components:
R1 250 V; R2 1000 V; C1 10 mF; C2 1 mF. (a) Bodediagram presenting frequency-dependent impedance magnitude
jZj together with the phase shift w of the sample under
investigation. (b) Wessel diagram locus of the same electrical
network. The imaginary component of the impedance
(reactance) is plotted against the real component (resistance).
The arrow indicates the direction along which the frequency
increases.
elements with frequency dependent properties. Such elements are not physically realizable with standard technical
elements. Table 1 provides a list of common circuit elements that are used to describe biomaterials with respect
to their impedance and phase shift. The constant phase
element (CPE) portrays a nonideal capacitor, and was
originally introduced to describe the interface impedance
of noble metal electrodes immersed in electrolytic solutions
(5). The physical basis for the CPE in living tissue (and at
electrode interfaces) is not clearly understood, and it is best
treated as an empirical element. Another example is the
Warburg impedance s that accounts for the diffusion limitation of electrochemical reactions (4).
It is important to place a word of caution concerning the
equivalent circuit modeling approach. Different equivalent
circuit models (deviating with respect to components or in
the network structure) may produce equally good fits to the
experimental data, although their interpretations are very
different. It may be tempting to increase the number of
elements in a model to get a better agreement between
experiment and model. However, it may then occur that the
model becomes redundant because the components cannot
be quantified independently. Thus, an overly complex
135
Figure 3. Calculated spectra of the

impedance magnitude jZj for the
electrical circuit shown in the
insert of Fig. 2b when the network
parameters have been varied
individually. The solid line in each
figure corresponds to the network
parameters: R1 250 V; R2
1000 V; C1 10 mF; C2 1 mF. (a)
Variation of C1: 10 mF (solid), 20 mF
(dashed), 50 mF (dotted). (b) Variation of C2: 1 mF (solid), 0.5 mF
(dashed), 0.1 mF (dotted). (c)
Variation of R1: 1000 V (solid),
2000 V (dashed), 5000 V (dotted).
(d) Variation of R2: 250 V (solid),
500 V (dashed), 1000 V (dotted).
model can provide artificially good fits to the impedance

data, while at the same time highly inaccurate values for
the parameters. Therefore, it is sound to use equivalent
circuits with a minimum number of elements that can
describe all aspects of the impedance spectrum (6).
Alternatively, the impedance data can be analyzed by
deriving the current distribution in the system with use of
differential equations and boundary values (e.g., the given
excitation at the electrode surfaces). The parameters of the
model impedances are then fitted to the data like described
above. An example of this approach is be presented below,
where it is used to analyze the IS of a cell-covered gold film
electrode.
IMPEDANCE ANALYSIS OF TISSUE AND SUSPENDED CELLS
The early and pioneering work on bioimpedance is associated with the names of Phillipson, et al. (5). In these
studies blood samples or pieces of tissue were examined in
an experimental setup as shown in Fig. 1, and the dielectric
properties of the biological system were investigated over a
broad range of frequencies from hertz to gigahertz.
To understand the origin of bioimpedance, it is necessary to look at the composition of living material. Any
tissue is composed of cells that are surrounded by an
extracellular fluid. The extracellular medium contains
proteins and polysaccharides that are suspended in an
ionic solution and the electrical properties of this fluid
are determined by the mobility and concentration of the

ions, primarily Na and Cl. The cell membrane marks the
boundary between the interior and exterior environment,
and consists of a 710 nm phospholipid bilayer. The membrane allows diffusion of water and small nonpolar molecules, while transport of ions and polar molecules requires
the presence of integral transport proteins. On the inside,
the cell contains a protein-rich fluid with specialized membrane-bound organelles, like the nucleus. For most purposes the fluid behaves as a pure ionic conductor. Thus, the
cell membrane is basically a thin dielectric sandwiched
between two conducting media and in a first approximation
its impedance characteristics are mainly capacitive.
The simplest possible explanatory model for biological
tissue (Fig. 4a-1) therefore consists of two membrane capa-
Table 1. Individual Impedance Contributions of Ideal and

Empirical Equivalent Circuit Elements
Component of
Equivalent
Circuit
Resistor
Capacitor
Coil
Constant phase
element (CPE)
Warburg impedance, s
Impedance,
Z
Phase
Shift, w
R
C
L
a(0 a 1)
R
(ivC)1
ivL
1/(iCv)a
0
p/2
p/2
ap/2
s(1 i)v0.5
p/4
Parameter
136
Figure 4. (a) Schematics of the

impedance measurement on living
tissue. The arrows indicate the
pathway of current flow for low
frequencies (solid line) and high
frequencies (dashed). Only at high
frequencies the current flows
through the cells. The electrical
structure of tissue can be directly
translated into equivalent circuit
1, which can be simplified to
equivalent circuit 2. (b) Bodediagram for network 2 in Fig. 4a,
using Rext 1000 V; Rcyt 600 V;
Cm 100 nF. (c) Impedance locus
generated with the same values.
citors Cm1 and Cm2 in series with a resistor of the cytosolic

medium Rcyt , which in turn acts in parallel with the
resistance of the conducting extracellular medium Rext .
Since the two series capacitances of the membrane cannot
be determined independently, they are combined to an
overall capacitance Cm . The equivalent circuit model of
this simple scenario (Fig. 4a-2) gives rise to a characteristic
impedance spectrum, as shown with the Bode-diagram
(Fig. 4b) and the Nyquist diagram (Fig. 4c). Impedance
data for biological tissue is also often modeled by the socalled ColeCole equation (7)
DR
Z R1
1 DRivCa
DR R0 R1
(4)
This simple empirical model is identical to the circuit of

Fig. 4, except that the capacitor is replaced by a CPE
element. The impedance spectrum is characterized by four
parameters DR; R1 ; a; t, where R0 ; R1 is the low- and
high frequency intercepts on the x axis in the Nyquist plot
(cf. Fig. 4d), t is the time constant t DR C, and a is the
CPE parameter. The impedance spectrum will be similar
to Fig. 4b, c, but when a 6 1, the semicircle in the Nyquist
diagram is centered below the real axis, and the arc will
appear flattened. For macroscopically heterogeneous
biological tissue, the transfer function is written as a
sum of ColeCole equations.
The features of the impedance spectrum Fig. 4b can be
intuitively understood: at low frequencies the capacitor
prevents current from flowing through Rcyt and the measured impedance arises from Rext . At high frequencies,
with the capacitor having a very low impedance, the current is free to flow through both Rcyt , Rext . Thus, there is a
transition from constant-level impedance at low frequencies to another constant level. This phenomenon is termed
dispersion, and will be discussed in the following.
A homogenous conducting material is characterized by a
bulk property named the resistivity r0 having the dimensions of ohm centimeters (V cm). Based on this intrinsic
parameter, the resistance may be defined by
R
r0 L
A
(5)
where A is the cross-sectional area and L is the length of

the material. Thus, by knowing the resistivity of the material and the dimensions of the system being studied, it is
possible to estimate the resistance. Similarly, a homogeneous dielectric material is characterized by an intrinsic
property called the relative permittivity e0 , and the capacitance is defined by
C
e0 e0 A
d
(6)
where e0 is the permittivity of free space with dimension

F/m, and A, d are the dimensions of the system as above.
For most biological membranes, the area-specific capacitance is found to be quite similar, with a value of 1 mF
cm2 (8).
For historical reasons the notation of conductivity s0
with dimensions S m1 and conductance (G s0 A=d) has
been preferred over resistance R and resistivity r, but the
information content is the same, it is just expressed in a
different way.
It is possible to recombine e0 and s0 by defining a complex
permittivity e e0 e00 , with Ree e0 and Ime e00 . The
137
imaginary part accounts for nonideal capacitive behavior,

for example, current within the dielectric due to bound
charges giving rise to a viscous energy loss (dielectric loss).
Therefore, e00 is proportional to s0 , when adjusted for the
conductivity that is due to migration s0 (9)
e00
s0 s0
2p f e0
(7)
When a piece of biological material is placed between two

electrodes, it is possible to measure the capacitance of the
system and thereby to estimated the tissue permittivity e0 .
In general, e0 quantifies the ratio of the capacitance when a
dielectric substance is placed between the electrodes, relative to the situation with vacuum in between. The increase
of capacitance upon insertion of a dielectric material is due
to polarization in the system in response to the electric
field. For direct current (dc) or low frequency situations e0 is
called the dielectric constant. When the frequency is
increased, e0 often shows strong frequency dependence
with a sigmoid character in a loglog plot of e0 versus
frequency. This step-like decrease of the permittivity is
referred to as a dielectric dispersion. The frequency fc at
which the transition is half-complete is called the characteristic frequency, and is often expressed as time constant t
with
1
t
(8)
fc
Going back to Fig. 4c, the characteristic frequency is found
directly as the point when the phase angle is at maximum.
The origin of dielectric dispersion in a homogeneous
material is due to a phenomenon termed orientation polarization. Dipolar species within the material are free to
move and orient themselves along the direction of the field,
and therefore they contribute to the total polarization.
However, when the frequency becomes too high, the dipoles
can no longer follow the oscillation of the field, and their
contribution vanishes. This relaxation causes the permittivity e0 to decrease.
For heterogeneous samples like tissue additional
relaxation phenomena occur, leading to more complex
frequency dependence. In 1957, Schwan (10) defined three
prominent dispersion regions of relevance for bioimpedance studies called a, b, and g, which is shown in
Fig. 5. The dispersions are generally found in all tissue,
although the time constant and the change in permittivity
De0 between the different regions may differ (9).
Briefly stated, the a-dispersion originates from the
cloud of counterions that are attracted by surface charges
of the cell membrane. The counterions can be moved by an
external electric field, thereby generating a dipole moment
and relaxation. The b-dispersion, which is also called
MaxwellWagner dispersion, is found in a window between
kilohertz and megahertz (kHz and MHz). It arises due to
the accumulation of charges at the interface between
hydrophobic cell membranes and electrolytic solutions.
Since the aqueous phase is a good conductor, whereas
the membrane is not, mobile charges accumulate and
charge up the membrane capacitor, thus, contributing to
polarization. When the frequency gets too high, the charging is not complete, causing a loss of polarization. Finally,
Figure 5. Frequency-dependent permittivity e0 of tissue. The

permittivity spectrum e0 (f) is characterized by three major
dispersions: a-, b-, and g-dispersion.
the g-dispersion is due to the orientation polarization of

small molecules, predominantly water molecules.
Most IS measurements are performed at intermediate
frequencies in the regime of the b-dispersion. In this
frequency window, the passive electrical properties of
tissue are well described with the simple circuit shown
in Fig. 4 or by the ColeCole equation. The measurements
can be used to extract information about extra- and intercellular resistance, and membrane capacitance. For example, it has been shown that cells in liver tissue swell, when
the blood supply ceases off (ischemia) and that the swelling
of the cells can be monitored as an increase in the resistance of the extracellular space Rext (11). Cell swelling
compresses the extracellular matrix around the cells,
and thereby narrows the ion pathway in this region. Based
on experiments like these, there is a good perspective and
prognosis that IS may serve as a routine monitoring tool for
tissue vitality even during the surgery.
APPLICATION: MONITORING OF ADHERENT CELLS
IN VITRO
The attachment and motility of anchorage dependent cell
cultures is conveniently studied using a microelectrode
setup. In this technique, cells are grown directly on a
surface containing two planar metal electrodes, one microelectrode and one much larger counter electrode. The cells
are cultured in normal tissue culture medium that serves
as the electrolyte.
When current flows between the two electrodes, the
current density, and the measured voltage drop, will be
much higher at the small electrode. Therefore the impedance measurement will be dominated by the electrode
polarization of the small electrode Zel . Instead, no significant polarization takes place at the larger counter electrode and its contribution to the measured impedance may
be ignored. The electrode polarization impedance Zel acts
physically in series with the resistance of the solution Rsol .
Since the current density is high in a zone (the constrictional zone) proximal to the microelectrode, the electrolytic resistance will be dominated by the constriction
resistance Rc in this region (Fig. 6). The total measured
impedance may therefore be approximated by Z Zel Rc
(4). If necessary, Rc may be determined from high frequency measurements where the electrode resistance is
138
Reference electrode
Tissue culture
medium
PC
Gold
electrodes
Cells
Photoresist
1 M
1V
Lock-in
amplifier
Figure 7. The ECIS measurement setup.
Rc
Constrictional
zone
Z pol
Microelectrode
Figure 6. Schematic of two-electrode configuration.
infinitely small, that is, ReZel ! 0, and subtracted from

the measured impedance to determine the impedance of
the electrodeelectrolyte interface.
When cells adhere on the small electrode, they constrain
the current flow from the interface, increasing the measured impedance. The changes mainly reflects the capacitive nature of nonconducting lipid-bilayer membrane
surrounding the cells. The cell membranes cause the current field to bend around the cells, much like if they were
microscopic insulating particles. It is possible to follow both
cell surface coverage and cell movements on the electrode,
and morphological changes caused by physiological/pathological conditions and events may be detected. The technique may also be used to estimate cell membrane
capacitances, and barrier resistance in epithelial cell
sheets. In addition, the method is highly susceptible to
vertical displacements of the cell body on the electrode with
sensitivity in the nanometer range.
INSTRUMENTATION
The technique was introduced by Giaever and Keese in
1984 and referred to as Electrical Cell-Substrate Impedance Sensing (ECIS) (12,13). The ECIS electrode array
consists of a microdisk electrode 5 104 cm2 and a
reference electrode 0:15 cm2 ; depending on the cell type
to be studied, the recording disk electrode may contain a
population of 20200 cells. The electrodes are made from
depositing gold film on a polycarbonate substrate over
which an insulating layer of photoresist is deposited and
delineated. A 1 V amplitude signal at fixed frequency (0.1
100 kHz) is applied to the electrodes through a large
resistor to create a controlled current of 1 mA, and the
corresponding voltage across the cell-covered electrodes
is measured by a lock-in amplifier, which allows amplification of the relatively small signals. The amplifier is interfaced to a PC for data storage. The impedance is calculated
from the measured voltage displayed in real time on the
computer screen (Fig. 7). During the measurements the
sample is placed in an incubator at physiological conditions.
The ECIS system is now commercially available, and
the electrode slides allow multiple experiments to be performed at the same time (14). Some modifications to the
technique have been described, such as a two-chamber
sample well, which permit simultaneous monitoring on a
set of empty electrodes being exposed to the same solution
(15), platinized single-cell electrodes (15), and inclusion of a
voltage divider technique to monitor the impedance across
a range of frequencies (16). More recently, impedance
studies have been performed using other types of electrode
design. One approach has been to insert a perforated
silicon-membrane between two platinium electrodes, there
by allowing for two separate electrolytic solutions to exist
on either side of the membrane (17). The results obtained
with these techniques are generally identical to those
obtained by the ECIS system.
MODEL OF ELECTRODECELL INTERFACE
To interpret ECIS-based impedance data, a model of the
ECIS electrodecell interface has been developed that
allows determination of (1) the distance between the ventral cell surface and the substratum, (2) the barrier resistance, and (3) the cell membrane capacitance of confluent
cell layers (18). The model treats the cells as disk shaped
objects with a radius rc that are separated an average
distance h from the substrate (Fig. 8). When cells cover
the electrode, the main part of the current will flow through
the thin layer of medium between the cell and the electrode, and leave the cell sheet in the narrow spacing
between cells. However, the cell membrane, which is modeled as a capacitor (an insulating layer separating the
conducting fluids of the solution and the cytosol) allows
a parallel current flow to pass through the cells. The minor
resistive component of the membrane impedance due to the
presence of ionic channels is ignored in the calculations. By
assuming that the electrode properties are not affected by
the presence of cells, a boundary-value model of the current
flow across the cell layer may be used to derive a relation
139
Cm
Rb
Ih
microelectrode
Figure 8. Model of current flow paths. The impedance changes
associated with the presence of cells, arise in three different
regions: from current flow under the cells quantified by a, from
current flow in the narrow intercellular spacings causing the
barrier resistance Rb. In parallel, some current will pass
through the cell membranes giving rise to capacitive reactance Cm.
between the specific impedance of a cell-covered electrode

Zcell and the empty electrode Zel
0
1
Zm
1
1 @ Zel
Zel Zm

A
Zcell Zel Z el Zm rc I0 r Rb 1 1

2 I1 r
Zel
Zm
s

1
1

9
h Zel Zm
where I0 ; I1 are the modified Bessel functions of the first
kind of order zero and one, Rb and r are the specific barrier
resistance and resistivity of the solution, and Zm
2i=vCm is the specific membrane impedance of the cells.
A parameter a rc r=h0:5 is introduced as an assessment
of the constraint of current flow under the cells. The
impedance spectrum of an empty electrode and a cellcovered electrode is used to fit the three adjustable parameters Rb ; a; Cm .
The model outlined above has been further refined to
describe polar epithelial cell sheets, treating separately the
capacitance of the apical, basal, and lateral membranes
(19). Some applications of the model will be discussed in the
following sections.
MONITORING ATTACHMENT AND SPREADING

As a cell comes into contact with a solid surface, it forms
focal contacts, primarily mediated by transmembrane
proteins that anchor structural filaments in the cell
interior to proteins on the substrate. During this process,
the rounded cell spreads out and flattens on the surface,
greatly increasing its surface area in contact with the
electrode. The cell will also establish contacts with neighboring cells through particular cellcell junctions, such as
tight junctions, where strands of transmembrane proteins
sew neighboring cells together, and gap junctions formed by
clusters of intercellular channels, connecting the cytosol of
adjacent cells.
The attachment process is normally studied using single-frequency measurements. Figure 9a and b show Bode
Figure 9. (a, b) Bode-diagrams of an ECIS electrode with

confluent MDCK cells and an empty electrode. (c) Plot showing
the division of the measured resistance of an ECIS electrode with
confluent MDCK cells with the corresponding values of the empty
electrode plotted versus log f.
plots for ECIS data of an empty electrode and an electrode

with a confluent layer of epithelial MDCK cells. It is seen
that the presence of cells primarily affects the impedance
spectrum for intermediate frequencies between 1 and 100
kHz (Fig. 9a). At the highest frequencies, the two plots
approach a horizontal plateau that represents the ohmic
solution resistance between the working and the counter
electrode. Within the relevant frequency window, the
140
phase-shift plot for the data of the cell-covered electrode

displays two extrema. At frequencies 200 Hz, the phase
shift w is closest to zero, indicating that the contribution of
the cells on the measured impedance is mainly resistive. At
higher frequencies, the effect of the cell layer becomes more
capacitive, and w starts approaching 908. The impedance
spectrum of the empty electrode displays a single dispersion related to double-layer capacitance at the electrode
interface.
The ideal measurement frequencies, where the presence
of cells is most pronounced, are determined by dividing the
impedance spectrum of a cell-covered electrode with the
spectrum of a naked electrode. The same can be done for
the resistance or capacitance spectrum, respectively. The
most sensitive frequency for resistance measurements is
typically found between 1 and 4 kHz (Fig. 9c), where the
ratio Rcell(f) / Rel(f) is at maximum. The capacitive contribution peaks at much larger frequencies, typically on the
order of 40 kHz, so that capacitance measurements are
often performed at this higher frequency.
During the initial hours following the initial electrode
cell contact, the monitored impedance undergoes a characteristic steep increase. Once the spreading is complete,
the impedance continues to fluctuate, reflecting the continuous morphological activities of living cells, for example,
movements of cells on the electrode, either by local protrusions or directed movements of the entire cell body, or cell
divisions (Fig. 10). The signal characteristics of the impedance during the spreading phase are generally found to be
distinct for different cell cultures, both in terms of the
duration of the initial gradient and its relative size in
comparison to the impedance recorded from a the naked
electrode (20). Also, characteristic impedance curves can be
obtained by coating the electrode with different proteins
(e.g., fibronectin, vitronectin) (21).
Simultaneous optical monitoring of a transparent ECIS
electrode has allowed systematic comparison of cell confluence and measured impedance (22). Analysis of data
from subconfluent MDCK epithelial cultures revealed a
strong linear association between the two variables with
cross-correlation coefficients > 0.9; the correlation was
found to be equally strong in early and late cultures. This
result indicates that 80% of the variance in the measured
20
Resistance (k)
Spreading complete
resistance (4 kHz) can be attributed to changes in cell

coverage area (Fig. 11). Moreover, it was possible to link
resistance variations to single-cell behavior during cell
attachment, including cell-division (temporary impedance
plateau) and membrane ruffling (impedance increase). The
measured cell confluence was compared to the theoretical
model (Eq. 9), neglecting the barrier resistance (i.e.,
Rb 0), and the calculated values were found to agree
well with the data (Fig. 12). Studies like these might pave
the way for standardized use of ECIS to quantify attachment and spreading of cell cultures.
Fluctuations
micromotions
IMPEDANCE SPECTROSCOPY AS A TRANDUCER

IN CELL-BASED DRUG SCREENING
15
BSC cells
Attachment
and
spreading
10
Spreading complete
NRK cells
No cells
Figure 11. Correlation between resistance and cell coverage. The

normalized resistance (4 kHz) versus time (upper panel), and the
electrode coverage versus time (lower panel) during the same time
interval. The measurement was started 32 h after the cells had
been seeded out; the cross-correlation factor was r 0.94.
Time (h)
10
15
Figure 10. Attachment assay of BSC and NRK fibroblastic cells

followed for an interval of 15 h. The graph shows the measured
resistance (4 kHz) as function of time; the spreading phase is
indicated with arrows.
Another application of impedance spectroscopy with strong

physiological and medical relevance is its use as transducer
in ECIS-like experiments for cell-based drug screening
assays. Here, the impedance readout can be used to monitor the response of cells upon exposure to a certain drug or a
drug mixture. In these bioelectric hybrid assays the cells
serve as the sensory elements and they determine the
specificity of the screening assay while the electrodes are
used as transducer to make the cell behavior observable. In
the following example, endothelial cells isolated from
bovine aorta (BAEC bovine aortic endothelial cells) were
grown to confluence on gold-film electrodes since they
141
Figure 12. Theoretical prediction of cell coverage. Theoretical

curve of normalized resistance plotted as function of cell coverage
on the electrode. Normalized resistance and corresponding cell
density are shown for four different registrations with circles.
Time points indicate when the recordings were initiated with
respect to the start of the culture; each circle corresponds to
average values for 15 min time intervals.
express cell-surface receptors (b-adrenoceptors) that are

specific for adrenalin and derivatives (23,24). These badrenoceptors belong to the huge family of G-protein
coupled receptors (GPCR) that are of great pharmaceutical
relevance and impact. By measuring the electrical impedance of the cell-covered electrode, the stimulation of the
cells by the synthetic adrenaline analogue isoprenaline
(ISO) can be followed noninvasively in real time without
any need to apply costly reagents or to sacrifice the culture
(25). Experimentally, the most sensitive frequency for
time-resolved impedance measurements is first determined from a complete impedance spectrum along an
extended frequency range as depicted in Fig. 13. The figure
compares the impedance spectrum of a circular gold-film
electrode (d 2 mm) with and without a confluent monolayer of BAECs. The contribution of the cell layer to the
total impedance of the system is most pronounced at
frequencies close to 10 kHz, which is, thus, the most
sensitive sampling frequency for this particular system.
It is noteworthy that the most sensitive frequency may
Figure 13. Frequency-dependent impedance magnitude for a

planar gold-film electrode (d 2 mm) with and without a
confluent monolayer of BAEC directly growing on the electrode
surface. The difference in impedance magnitude is maximum at a
frequency of 10 kHz.
Figure 14. (a) Time course of the impedance magnitude at a

sampling frequency of 10 kHz when a confluent monolayer of
BAECs is exposed to 10 mM ISO or a corresponding vehicle
control. (b) Dose-response relationship between the increase of
impedance magnitude DZ and the concentration of isoproterenol
applied. Quantitative analysis reveals an EC50 of 0.3 mM similar to
the binding constant of ISO to b-adrenoceptors.
change with the electrode size and the individual electrical

properties of the cells under study.
Figure 14a traces the time course of the impedance
magnitude at a frequency of 10 kHz when confluent BAEC
monolayers were either challenged with 10 mM ISO or a
vehicle control solution at the time indicated by the arrow.
The exchange of fluids produces a transient rise of the
impedance by 1020 V that is not caused by any cellular
response, but mirrors the reduced fluid height within the
measuring chamber. As expected, no response of the cells is
seen in the control experiment. The cell population exposed
to 10 mM of ISO shows a significant increase in electrical
impedance that goes through a maximum 10 min after ISO
application, and then slowly declines. The reason for the
increase in impedance as observed after ISO stimulation is
similar to what has been described for three-dimensional
(3D) tissues above. The adrenaline derivative induces a
relaxation of the cytoskeleton that in turn makes the cells
flat out a bit more. As a consequence the extracellular space
between adjacent cells narrows and increases the impedance of the cell layer. Note that the time resolution in
these measurements is 1 s so that even much faster cell
responses than the one studied here can be monitored in
real time. Moreover, no labeled probe had to be applied and
142
the sensing voltages used for the measurement (U0 10

mV) are clearly noninvasive.
From varying the ISO concentration, a dose-response
relationship (Fig. 14b) can be established which is similar
to those derived from binding studies using radiolabeled
ligands. Fitting a dose-response transfer function to the
recorded data returns the concentration of half-maximum
efficiency EC50 as (0.3 0.1) mM, which is in close agreement to the binding constant of ISO to b-adrenoceptors on
the BAEC surface as determined from binding assays with
radiolabeled analogs (23).
These kind of electrochemical impedance measurements are also used to screen for potent inhibitors of
cell-surface receptors. Staying with the example discussed
in the preceding paragraph, the blocking effect of Alprenolol (ALP), a competitive inhibitor of b-adrenoceptors
(b-blocker), is demonstrated. Preincubation of BAEC with
ALP blocks the stimulating activity of ISO, as shown in
Fig. 15. The figure compares the time course of the impedance magnitude at a frequency of 10 kHz when BAEC
monolayers were stimulated with 1 mM ISO either in
absence of the b-blocker (a) or after preincubation (b).
Figure 15. (a) Time course of the impedance magnitude at a

sampling frequency of 10 kHz, when confluent BAEC are
exposed to 1 mM ISO. (b) Time course of the impedance
magnitude of a confluent monolayer of BAECs upon sequential
exposure to 10 mM of the b-blocker ALP and 1 mM ISO 20 min later.
The b-adrenergic impedance increase is omitted by the b-blocker.
Intactness of the signal transduction cascade is verified by
addition of forskolin (FOR), a receptor independent activator of
this signal transduction pathway.
Figure 16. Time course of the impedance magnitude at a

sampling frequency of 10 kHz when a confluent BAEC
monolayer is exposed to an over dose of the b-blocker alprenolol
(100 mM ALP). Addition of ALP is indicated by the arrow.
When the cell layers were incubated with 10 mM ALP prior

to the addition of 1 mM ISO, the cells do not show any ISO
response indicating that a 10-fold increase of ALP was
sufficient to block activation of the receptors. To prove that
the cells were not compromised in any way during these
experiments the same signal transduction cascade was
triggered via a receptor-independent way at the end of
each experiment. This can be easily done by application of
FOR, a membrane permeable drug that intracellularly
activates the same enzyme that is triggered by ISO binding
to the receptor. Forskolin stimulation of those cells that
had been blocked with ALP earlier in the experiment
induces a strong increase of electrical impedance indicating that the intracellular transduction pathways are functional (Fig. 15b).
Besides screening for the activity of drugs in cell-based
assays, these kind of measurements are also used to check
for unspecific side effects of the compounds of interest on
cell physiology. Dosage of ALP and many of its fellow bblockers has to be adjusted with great care since these
lipophilic compounds are known to integrate nonspecifically into the plasma membrane. As shown in Fig. 15b,
application of 10 mM ALP does not show any measurable
side effects. Using ALP in concentrations of 100 mM induces
a transient but very pronounced reduction of the electrical
impedance (Fig. 16). This decrease in impedance may be the
result of the interaction of ALP with the plasma membrane
or an induced contraction of the cell bodies.
The preceding examples showed that impedance measurements of cell-covered gold electrodes in which the cells
serve as sensory elements can be used in screening assays
for pharmaceutical compounds, but also for cytotoxicity
screening. The interested reader is referred to Ref. 26
and 27.
SUMMARY AND OUTLOOK
Impedance spectroscopy is a general technique with important applications in biomedical research and medical
diagnostic practice. Many new applications are currently
under investigation and development. The potential of the
technique is obviously great, since it is noninvasive, easily

applied, and allows on-line monitoring, while requiring
low cost instrumentation. However, there are also difficulties and obstacles related to the use of IS. Foremost,
there is no separate access to the individual processes and
components of the biological system, only the total impedance is measured, and this signal must be interpreted by
some chosen model. There are many fundamental issues
yet to be solved, both connected with understanding the
origin of bioimpedance, methodological problems with
finding standardized ways of comparing different samples, as well as technical issues connected with the equipment used to probe bioimpedance.
Prospective future in vivo applications include quantification of ischemia damage during cardiac surgery (28)
and organ transplantation (29), as well as graft rejection
monitoring (30). Impedance spectroscopy are also used
for tissue characterization, and recently a device for
breast cancer screening became commercially available.
Multifrequency electrical impedance tomography (EIT)
performing spatially resolved IS is a potential candidate
for diagnostic imaging devices (31), but due to poor resolution power compared to conventional methods like MR,
only few clinical applications are described.
The use of impedimetric biosensor techniques for in
vitro monitoring of cell and tissue culture is promising.
With these methods, high sensitivity measurements of cell
reactions in response to various stimuli have been realized,
and monitoring of physiologicalpathological events is
possible without use of marker substances. The potential
applications cover pharmaceutical screening, monitoring
of toxic agents, and functional monitoring of food additives.
Microelectrode-based IS is interesting also for scientific
reasons since it allows studying the interface between cells
and technical transducers and supports the development of
implants and new sensor devices (32).
Finally, affinity-based impedimetric biosensors represent an interesting and active research field (33) with
many potential applications, for example, immunosensors
monitoring impedance changes in response to antibody
antigen reactions taking place on electrode surfaces.
BIBLIOGRAPHY
1. Macdonald JR. Impedance Spectroscopy. New York: John
Wiley & Sons; 1987.
2. Fricke H, Morse S. The electrical capacity of tumors of the
breast. J Cancer Res 1926;10:340376.
3. Schwan H. Mechanisms Responsible for Electrical Properties
of Tissues and Cell Suspensions. Med Prog Technol 1993;19:
163165.
4. Grimnes S, Martinsen G. Bioimpedance and Bioelectricity
basics. Cornwall: Academic Press; 2000.
5. McAdams E, Lackermeier A, McLaughlin J, Macken D, Jossinet J. The linear and non-linear electrical properties of the
electrode-electrolyte-interface. Biosens Bioelectron 1995;10:
6774.
6. Kottra G, Fromter E. Rapid determination of intraepithelial
resistance barriers by alternating current spectroscopy. II.
Test of model circuits and quantification of results. Pflugers
Arch 1984;402:421432.
143
7. Cole K, Cole R. Dispersion and adsorption in dielectrics. I..

alternating current characteristics. J Chem Phys 1941;9:
341351.
8. Cole Ks. Membrane, Ions and Impulses. Berkeley (CA): University of California Press; 1972.
9. Kell D. Biosensor. Fundamentals and Applications. Turner A,
Karube I, Wilson G, editors. Oxford Science Publications;
1987. pp 427468.
10. Schwan H. Electrical properties of tissue and cell suspensions.
Advances in biological and medical physics. Lawrence J,
Tobias C, editors. New York: Academic Press; 1957. pp 147
209.
11. Gersing E. Impedance spectroscopy on living tissues for determination of the state of organs. Bioelectrochem Bioenenerg
1998;45:149.
12. Giaever I, Keese CR. Monitoring fibroblast behavior in tissue
culture with an applied electric field. Proc Natl Acad Sci
1984;81:37613764.
13. Giaever I, Keese CR. A morphological biosensor for mammalian cells. Nature (London) 1993;366:591592.
14. Applied Biophysics, Inc. 2002.
15. Connolly P, et al. Extracellular electrodes for monitoring cell
cultures. IOP Publishing; 1989.
16. Wegener J, Sieber M, Galla HJ. Impedance analysis of epithelial and endothelial cell monolayers cultured on gold surfaces.
J Biochem Biophys Methods 1996;76:327330.
17. Hagedorn R, et al. Characterization of cell movement by
impedance measurements on fibroblasts grown on perforated
Si-membranes. Biochem Biophys ActaMolecular Cell Res
1955;1269:221232.
18. Giaever I, Keese C. Micromotion of mammalian cells measured electrically. Proc Natl Acad Sci USA 1991;88:7896
7900.
19. Lo CM, Keese CR, Giaever I. Impedance analysis of MDCK
cells measured by electric cell-substrate impedance sensing.
Biophys J 1995;69:28002807.
20. Giaever I, Keese CR. Use of electric fields to monitor the
dynamical aspect of cell behavior in tissue culture. IEEE
21. Mitra P, Keese CR, Giaever I. Electrical measurements can be
used to monitor the attachment and spreading of cells in tissue
culture. BioTechniques 1991;11:504510.
22. De Blasio BF, Laane M, Walmann T, Giaever I. Combining
optical and electrical impedance techniques for quantitative
measurements of confluence in MDCK-I cell cultures. BioTechniques 2004;36:650662.
23. Zink S, Roesen P, Sackmann B, Lemoine H. Regulation of
endothelial permeability by beta-adrenoceptor agonists: Contribution of beta 1- and beta 2-adrenoceptors. Biochim Biophys
Acta 1993;1178:286298.
24. Zink S, Roesen P, Lemoine H. Micro- and macrovascular
endothelial cells in beta-adrenergic regulation of transendothelial permeability. Am J Physiol 1995;269:C1209
C1218.
25. Wegener J, Zink S, Roesen P, Galla H. Use of electrochemical
impedance measurements to monitor beta- adrenergic stimulation of bovine aortic endothelial cells. Pflugers Arch
1999;437:925934.
26. Arndt S, et al. Bioelectrical impedance assay to monitor
changes in cell shape during apoptosis. Biosens Bioelectron
2004;19:583594.
27. Keese C, Karra N, Dillon B, Goldberg A, Giaever I. Cellsubstratum interactions as a predictor of cytotoxity. In Vitro
Mol Toxicol 1998;11:183191.
28. Benvenuto, et al. Impedance microprobes for myocardial
ischemia monitoring. 1st Annual International IEEE-EMBS.
Lyon, France; 2000. p 234238.
144
INCUBATORS, INFANTS
29. Haemmerich D, et al. Changes in electrical resistivity of

swine liver after occlusion and postmortem. Med Biol Eng
Comput 2002;40:2933.
30. Ollmar S. Noninvasive monitoring of transplanted kidneys by
impedance spectroscopya pilot study. Med Biol Eng Comput
1997;35:1336.
31. Brown B. Electrical impedance tomography (EIT): A review. J
Med Eng Technol 2003;27:97108.
32. Borkholder D. Cell based biosensors using microelectrodes.
Ph.D. Dissertation. 1998. Stanford University.
33. Katz E, Wilner I. Probing biomolecular interactions at conducting and semiconducting surfaces by impedance spectroscopy: routes to impedimetric immunosensors. Electroanalysis
2003;15:913947.
See also CAPACITIVE
MICROSENSORS FOR BIOMEDICAL APPLICATIONS.
IMPLANT, COCHLEAR.
See COCHLEAR
PROSTHESES.
INCUBATORS, INFANTS
ROBERT WHITE
Memorial Hospital South Bend
South Bend, Indiana
INTRODUCTION
Providing newborn infants with appropriate thermal protection is known to improve their growth rates (13),
resistance to disease (4,5) and survival (611). Keeping
premature, sick, or otherwise high risk babies warm is
particularly critical and, when their care precludes covering them with protective swaddling clothing, especially
difficult. Incubators are devices used during the care of
such high-risk infants and are designed with the intent of
producing environmental conditions that are consistently
suitable to each unique infants particular needs. There are
many different kinds of incubators that differ in detail in
the way they are constructed, heated, and controlled
(1216). All provide a mattress for the infant to lie upon,
surrounded by a warmed microclimate that is controlled by
a logical system governing the amount of heat needed to
keep the environmental temperature within a limited
range. In some incubators, this microclimate is produced
within a rigid walled chamber; such devices are called
closed incubators. When they are heated by using a fan
to force air over a metallic heating coil prior to its entry into
the infants chamber, these closed incubators are also
called forced convection incubators. There also are open
incubators; those have no walls and, therefore, no chamber
surrounding the mattress. There is nothing delimiting the
convective environment in an open device, so they need to
be heated by using a radiant warmer directed to the
mattress area. These devices, therefore, are commonly
called open radiant incubators, radiant warmer beds, or
radiant heaters.
Each of these types of incubators provides certain
unique advantages. The convectively heated incubator
provides a caretaker with a far easier method for controlling the humidification of the infants microclimate, when
compared to the open radiant warmer bed. Therefore, a
baby under an open radiant heater loses more body fluid
than does an infant within a closed convectively heated
chamber (17). But conversely, a baby in an open incubator,
while more complicated to care for in terms of medical
fluids administration, is physically more accessible in
terms of other kinds of care that sometimes are equally
important to the well being of sick babies. Current top of
the line incubators incorporate the advantages of both
types, utilizing a radiant warmer bed with a removable
enclosure that allows full physical access to the infant
when the incubator is operated in the radiant heater mode,
and better control of humidification and noise when the
enclosure is placed around the baby and operated in the
convectively heated mode.
An incubator, in many respects, is just a very little
house sized to fit the space and functional requirements
of an infant occupant. As choices must be made when
conditioning the environment in any house, different
options must be considered when designing the climate
control system in an incubator. In the following review,
some of these considerations will be explained from the
perspective of how environmental manipulators affect
newborn infants who are not just little human adults,
but also developing individuals with special physical,
physiologic, metabolic, and neurological capabilities
and limitations that make them unique. In great measure
incubator manufacturers have been successful in translating present day knowledge of babies and their special needs
into technical solutions that make todays incubators
remarkably functional. But any infant caretaker or incubator manufacturer can attest to the limitations of todays
devices which, as they are approximate to our present
scientific knowledge and the existing level of technology,
are flawed by our considerable remnant ignorance and the
failure of existing technology to meet certain imperative
needs already known.
HISTORY
It is ancient knowledge that infants who are allowed to get
cold have a greater chance of dying than do infants kept
warm. Prior to the nineteenth century, keeping small
babies warm meant swaddling with multiple layers of
cloth, providing body contact with the mother, or placement of the infant near a warm, roaring fireplace. Such
classic thermal care served lusty, healthy babies well, but
was inadequate to provide for the special needs of premature or otherwise enfeebled newborns. These special needs
were not met because, until the last century, there was
almost no recognizable major medical or social commitment toward enhancing the survival of babies born prematurely. The infant mortality rate was high and accepted.
However, in response to various politicosocial events that
occurred in the late 1700s and early 1800s, the value of
improving premature infant survival increased, stimulating the development of special incubators in which to care
for these fragile, newly valued babies.
INCUBATORS, INFANTS
145
Figure 1. von Ruehl warming tub (1835). Reprinted with permission from T. E. Cone, Jr., History of the Care and Feeding of the
Premature Infant, Boston, MA: Little, Brown, 1985.
The first serious attempt to improve on the thermal

protection provided newborns was reflected in a warming
tub developed in 1835 by Johann Georg von Ruehl
(17691846) in Russia (Fig. 1). The von Ruehl tub was
simply a double-walled sheet-iron open cradle that was
kept warm by filling the space between the walls with
warm water. Variations on von Ruehls design were subsequently developed throughout Europe, and this type of
primitive open incubator remained a standard device for
care until 1878.
Although the von Ruehl device must be recognized as a
developmental milestone, to be truly accurate, the developmental history of modern infant incubators must be
traced back centuries to Egypt where the artificial incubation of eggs was refined and remained a closely guarded
secret and uniquely Egyptian profession. Not until 1799
were these secrets introduced into Europe by members
of Napoleons expedition. Professor Stephane Tarnier
(18281897) of the Paris Maternity Hospital in 1878 saw
a chicken incubator at the Paris Zoo. The incubator, based
on old Egyptian designs, had been constructed by Odile
Martin. Dr. Tarnier perceived how such a device, with
modifications, could be used to keep premature infants
warm. Odile Martin subsequently built the first approximation of the modern enclosed infant incubator initially
used at the Paris Maternity Hospital in 1880 (Fig. 2)
The Tarnier incubator was simple in its design. The
infant lay in the upper chamber of a two-chambered
double-walled box insulated to slow the loss of heat. The
infant chamber was topped with a removable cover through
which the infant could be observed while remaining protected from cooling room drafts. The heating of the upper
chamber was achieved by warming a large supply of water
contained in the lower chamber of the incubator. The water
was heated by an alcohol or gas lamp thermosyphon that
was external to the incubator chambers and connected by
piping that allowed convection driven water flow between
the heater and the water reservoir. Cool room air freely
flowed into the lower warming chamber where the air
picked up heat from the surface of the warm water reservoir and then, by natural convection, rose to enter and
warm the upper chamber containing the infant.
The Tarnier incubator was neither elegant nor efficient,
and even when within the device, infants needed the extra
Figure 2. Tarnier incubator (1880). Reprinted with permission

from T. E. Cone, Jr., History of the Care and Feeding of the
Premature Infant, Boston, MA: Little, Brown, 1985.
protection of swaddling blankets. It did, however, reflect

the technology of its day and, in the climate of a new
commitment toward the study and welfare of feeble
infants, stimulated others to refine the basic Tarnier
design to correct deficiencies discovered through acquired
practical experience with the incubator in clinical settings.
The historical progression in this development has been
illuminated by Dr. Thomas Cone, and the reader is referred
to Dr. Cones excellent treatise for a more detailed description of the many design variations introduced and tested
along this early path leading to todays equipment (18).
The modernization of incubators has not only led to
marked improvement in the thermal protection provided
infants, but it also has been pivotal in increasing our
knowledge of diseases unique to newborn babies. In turn,
each increment of knowledge has required manufacturers
to modify incubators in order to provide solutions to
problems the new scientific discovery imposed. For example, when electric fans became available and forced air
convection systems were developed, incubator heating
became so improved that infants for the first time could
be undressed during care. This along with the use of new
transparent plastics in the construction of incubator walls
allowed clinicians to make observations that led to detailed
descriptions of illnesses about which little was known
when infants were hidden within the predominately opaque wooden and metal chambers of the past. But while the
employment of clear plastic in incubator construction
enhanced the ability to observe infants, its poor insulating
qualities made the task of maintaining the incubator
chamber in a warm and stable state more difficult. And
as improved visibility led to new lifesaving therapies, such
as the administration of intravenous fluids, the use of
respirators in care, and the development of new diagnostic
procedures and surgical interventions, the transparent
plastic walls that provided caretakers with visual access
to sick babies during these therapeutic processes also
served as impediments. Incubators had to be modified so
that catheters, tubes, and wires could be connected to an
infants body. Increasing numbers of access holes were
146
INCUBATORS, INFANTS
drilled through the walls of the incubator to provide portals

of entry for these therapeutic tools, but the new fenestrations also produced new exits for life-sustaining heat. More
modifications were needed and, as each problem was
solved, a new dilemma emerged.
Even today, infant caretakers and incubator manufacturers continue to struggle with these and other problems
contributing to the strengths and weaknesses in incubator
devices. In this article, the physiologic, clinical, and technical factors leading to the design of existing incubators
will be outlined further and some of the limitations of
incubators explained in greater detail. Throughout, we
hope that it remains clear that incubator development is
an ongoing process requiring frequent and critical review
of existing methods to assure that the thermal protection
being provided is still appropriate during the delivery of
other and especially newer forms of care also deemed
necessary to a babys well being. The ideal incubator of
tomorrow is one that neither impedes care nor can itself,
when providing thermal protection, be impeded by other
forms of care. This has always been and remains the major
challenge to health care providers and engineers committed to incubator development.
FACTORS CONSIDERED IN INCUBATOR DESIGN
Physiological Heat Balance
Even though some controversy exists concerning the exact
definition of the body temperature limits within which a
newborns body functions optimally, in general, any temperature between 35.5 and 37.5 8C is probably within that
normal range and can be referenced to published data
available on the subject. Body temperature is determined
by the balance between the heat produced within and lost
from the body tissues. In order to design or even understand the design and limitations of modern incubators, a
knowledge of these basic factors is required to provide the
context for how infants differ from adults in their thermoregulatory capabilities.
Heat Production
All animals, including human babies, produce body heat as
a by-product of the biochemical processes that sustain life.
The basic amount of heat produced by a newborn infant is
1.52 W kg1. During the first weeks of life, this minimal
rate of heat production is both weight and age related, with
smaller and younger babies capable of producing less heat
than larger and older infants (1923).
In addition to this basic capacity, most healthy babies
have the capability to generate additional heat to a
maximum production rate of 4.55 W kg1 (2123). This
additional heat-producing capacity is often called upon for
protective purposes, as, for example, when the infant is
challenged to fight off infection or when stressed by situations that cause an exorbitant amount of heat to be lost
from the body. The capability to increase the amount of
heat produced to replace body heat losses is called homeothermy. In contrast to homeotherms, some creatures,
such as lizards, reptiles, and fish, are poikilotherms that
do not produce more heat when cooled, but actually

decrease their metabolic rates when exposed to cold.
When considering thermoregulatory problems associated with newborn care, both homeothermy and poikilothermy must be understood, because under some
circumstances, it is possible for a homeothermic animal
to behave like a poikilotherm. Sometimes during the medical care of humans this possibility is used to a patients
benefit; for example, during some heart operations patients
are given drugs that prevent their nervous systems from
responding to the cold. In this circumstance, it is desirable
to slow the bodys metabolic rate, which can be achieved by
cooling the drug treated patient who, by intent, has been
changed to a temporary poikilotherm.
At times, a homeotherm may revert temporarily to a
poikilothermic state. This is particularly common in immature or very sick newborns, and especially those with
neurologic damage (24) or with breathing problems that
lead to inadequate blood and tissue oxygen levels (25,26).
Poikilothermy in a newborn can also occur because of drugs
administered to a mother in labor with subsequent placental transport of those drugs to the infant (27).
In spite of their occasional reversion to poikilothermy, it
nonetheless is commonly advised that newborns should be
thought of as homeotherms and protected from environments that would unduly stimulate homeothermic tendencies. This is because homeotherms increase heat
production by increasing metabolic work which can cause
excess utilization of finite fat, sugar, and protein stores
needed to sustain other vital body functions and to meet
growth and developmental milestones. Moreover, extra
metabolic work produces more than just extra heat; acidic
and other metabolic by-products produced at the same time
can cause severe imbalances in the bodys critical acidbase
balance (4). As a consequence of the self-protective reaction
to cold stress a newborn may, therefore, be faced with an
equally life-threatening biochemical stress (Fig. 3).
It has been suggested that one reason cold-exposed
infants have higher mortality rates is that they become
metabolically exhausted and incapable, in part because of
the consequent acidosis, to fight off the stresses placed on
their bodies by other threatening events. But problems can
arise when attempts are made to provide infants with
protection from cold stress, since it is unclear how to be
sure that a given environment minimizes the infants
metabolic efforts to maintain homeothermy. Theoretically,
this could be achieved by measuring the infants metabolic
rate continuously and making adjustments in the environmental supports whenever the infants rate of metabolism
changed, but measurement of heat production by a newborn is very difficult in the clinical setting.
In any case, few infant caretakers would consider the
infants metabolic rate as the only measure of success when
providing a baby with thermal protection. The infants
actual body temperature is also considered important
(21) and it is well known that metabolic rate is only one
factor contributing to the body temperature measured.
Body temperature also is influenced by the rate at which
heat is lost from the body and, if one wishes to produce a
device that contributes to the maintenance of an infants
body temperature, it is necessary, by virtue of its balancing
INCUBATORS, INFANTS
147
as infants transpire water from their skin into the surrounding environment. They can also lose heat by evaporation as residual amniotic fluid or bath water dries from
their skin and hair, and they lose heat from their lungs as
they exhale warm humid air.
Gross estimates of the magnitude of heat loss can be
calculated by physical heat-transfer equations for each
mechanism. The reader is referred to thermal transfer
books for the details of these mechanisms, but examination
of these equations here in a simplified form is a useful way
to discover some of the special features that influence heat
transfer as applied to newborn care.
All nonevaporative heat losses are quantitatively proportional to the magnitude of the temperature difference
between the warmer object (To) losing heat and the cooler
environmental feature (Te) that will receive the heat in
transfer:
Heat loss / T0 Te
This equation becomes equality by adding to it an object
specific constant called a thermal transfer coefficient (k):
Figure 3. Schematic of homeothermy in newborns. On sensing
loss of body heat, the infant minimizes heat loss from the skin by
vasoconstricting blood vessels, changing body positions, and
increasing metabolic rate. The increase in metabolism can
produce acidosis and depletion of energy substrate stores.
Reproduced by permission from P. H. Perlstein, Routine and
special carePhysical environment. In A. A. Fanaroff and R. J.
Martin (Eds.), Behrmans Neonatal-Perinatal Medicine, 3rd ed.,
St. Louis, MO: C. V. Mosby Co., 1983.
effect on the final body temperature achieved, to understand how heat loss occurs in newborns.
Heat Loss
In the final analysis, incubators actually protect infants
only by modifying the factors that contribute to heat loss
via the well-known conductive, convective, radiant, and
evaporative mechanisms.
The flow of heat occurs only when there is a difference in
the temperatures of adjacent structures. Heat can only be
lost by a warmer object to a cooler one. Conductive heat
losses occur when an infant comes in physical contact with
a cooler solid surface. A baby loses heat by conduction to a
cooler mattress, blanket, diaper, or other clothing.
Convective losses are similar to, but independent of, conductive losses and occur when a baby is exposed to air
currents that are cooler than the infant. Convective losses
are influenced not only by temperature differences, but
also by the wind chill factor determined by speed at which
the air is flowing. In a practical sense, this means that if an
incubator has a fan as part of a forced convection heating
system, air movement produced by the fan can cause a
cooling effect in excess that which would occur if air at the
same temperature was still.
Heat loss in infants also occurs by radiation in the
infrared (IR) spectrum to cooler solid objects surrounding,
but not in contact with their skin. The incubator and room
walls, windows, and furniture all contribute to heat loss via
radiant mechanisms. Finally, evaporative heat losses occur
Heat loss kT0 Te

Different materials at the same temperature lose heat
at different rates when exposed at the same thermal environment; for example, a block of wood has a lower thermal
transfer coefficient than a block of steel. Newborn infants
have higher thermal transfer coefficients than do adults
and therefore lose body heat more rapidly than adults
when exposed to any environment that is cooler than body
temperature, so in a room temperature that feels comfortable to an adult, a newborn can get cold.
It must be emphasized that the heat loss equation as
written above is grossly simplified and omits numerous
other factors important to actual heat exchange. A more
accurate equation would have to include a factor accounting for the infants exposed surface area, which is a quantity that changes with changes in an infants position and is
modified if the infant is swaddled in blankets, wears a
diaper, booties, hat, or, if in the course of surgical care, has
a bandage applied. The equation as simplified particularly
fails to reflect the true degree of complexity describing the
thermal relationship between an infants skin and the
radiant surfaces of a room or incubator. The relationship
is modified, for example, by complex factors describing the
exact path and distance traveled by infrared (IR) waves in
their transfer from object to object.
Radiant heat loss is also modified by the emissivities of
the objects exchanging energy. Like black carbon, an
infants skin, no matter what its actual color, is presumed
to have an emissivity of 1, which means it absorbs and
emits IR rays perfectly and completely. The emissivities of
the materials used in incubator manufacture or in nursery
wall coverings also modify the amount of radiant exchange
occurring with the infants radiant surface. The emissivities of these objects become particularly important in an
incubator chamber in which the surface area of the interior
walls surrounds the exposed surface area of the infants
radiating skin.
The following equation, although still simplified, provides a better approximation of expected radiant losses (Hr)
148
INCUBATORS, INFANTS
from an infant in an incubator (28). In this equation, Ab is

the exposed surface area of the infant, Ar is the area of the
walls surrounding the infant, Es is the emissivity of
the infants skin, and Er is the emissivity of the walls.
The symbol o is the StefanBoltzmann constant, 5.67
108 W1 m2 K4. When using this equation, temperatures are expressed in kelvin and the heat loss in watts.

1
1
Ab 1
H r Ab
1
sTs4 Tr4
Es Ar Er
Radiant exchange relationships are so variable because
of differences between infants and different incubator
environments that, even using this more complex equation,
only poor and static quantitative approximation of actual
radiant flux can be made in clinical settings. This proves to
be a practical problem when considering incubator designs,
since it has been documented that in many situations
radiant losses can account for >60% of the total heat loss
from an infant (29).
Evaporative heat losses are not specifically related to
temperature difference and occur instead because of differences that exist between the partial pressures of water in
the boundary layer of air next to the infants skin and that
in the environment beyond the boundary layer limits.
Evap loss Kpartial pressure skin
partial pressure airares
Partial pressures are not the same as relative humidities, so even in an environment at 100% relative humidity,
an infant can lose water and heat if the skin surface is
warmer than the environment. For each milliliter of fluid
lost from the body, 580 g cal (2.4 kJ) of heat are lost in the
vaporization process. This route of heat loss accounts for
25% of the total heat loss when an infant is dry. When
lying unclothed on an open bed heated only by a radiant
heater, up to 300 mL1 kg1 day1 of fluid can be lost by
evaporation from the skin of very immature infants in the
first days of life. In an enclosed incubator that is poorly
humidified, up to 150 mL1 kg1 day1 of water can be lost
by this mechanism in very immature infants. Following
birth when the infant is wet with amniotic fluid, or following a bath, this can become the predominant route of heat
loss (3034,34).
Environmental Temperature
It should be obvious from the previous discussion that since
conduction, convection, radiation, and evaporation are
each relatively independent mechanisms, no single measurable quantity can be used to calculate their combined
contribution to heat loss. Air temperature, for example, can
be used to estimate only the convective component of heat
loss from a baby, while measurements of incubator inside
wall temperatures can only be helpful in determining
approximate losses due to radiation. This means that if
the incubator walls are cold, a baby in an incubator can lose
heat even if the air temperature is warmer than the infant.
The only feature necessary for this to be true is for radiant
losses to be higher than convective heat gains.
Environmental temperature, although frequently used
loosely to describe any ambient thermal value, must be
understood to be a specific reference to the combination of

temperatures actually experienced by an infant in thermal
exchange relationships via multiple mechanisms. Unfortunately, few guidelines exist at present to help caretakers
know the true environmental temperature for an infant
within an incubator in a clinical setting. When certain
conditions are met, however, some of the guidelines seem
to be useful. Dr. Hey, for example, determined that in a
well-humidified enclosed convectively heated incubator
with single-layer Plexiglas walls, the environmental temperature perceived by a contained infant is 1 8C lower
than the measured midincubator chamber air temperature
for every 7 8C difference that exists between the incubator
air temperature and the air temperature of the room
within which the incubator stands (35).
Heat Transfer within the Body
Since the skin of the newborn is the major heat-losing
surface in exchange with the environment, mechanisms
by which heat transfers from interior tissues to the skin
play an important part in the heat loss process. The rate at
which internally produced heat is transferred from the
body core temperature TB through body tissues to the outer
body skin surface at temperatures Ts is computed using the
following equation:
Heat transfer CTB Ts
Where C is an individuals specific thermal conductance
coefficient, which is affected by the absolute thickness and
character of the skin, subcutaneous fat, and other subcutaneous tissue, and by the blood flow rate from the body
core to its surface. Obviously, babies are smaller and have
thinner body coverings than do adults, and, therefore, as
they lose heat more rapidly from their surfaces than do
adults, they also transfer heat more rapidly to their surfaces from their internal heat-producing organs. In addition, an infants blood vessels are relatively close to the
body surface. Since the vascularity of a particular body
surface determines the rate at which blood will shunt core
heat around intervening insulating tissues to the skin
surface, such shunting contributes heavily to the high
thermal conductance of a baby.
Heat can also be lost rapidly from an infants body core
via the respiratory system. This route of loss is of relatively
minor significance in a healthy and spontaneous breathing
infant, but in a baby who is ill and especially one being
mechanically assisted by a respirator, this can become the
major route by which body heat is lost or gained. Body heat
transferred by this route is dependent on the infants
temperature, breathing rate, the flow rate and temperature of gases reaching the lungs, and the water content of
the gas delivered to the airway. If temperatures and humidification are not properly monitored and controlled, the
heat losses from the respiratory passages can be so great
that they exceed the capacity of the incubator heater.
The Concept of a Neutral Thermal Environment
Theoretically and as demonstrated by several authors
(21,36,37), it is possible for a competent homeothermic
baby to have a body temperature that is below the normal
INCUBATORS, INFANTS
range at a time when the measured metabolic rate of the

infant is at a minimal unstimulated level. For example,
Bru ck (36) documented that during a period of cooling
associated with a falling environmental temperature, an
infants metabolic rate and heat production increased, but,
as soon as the cooling environment was caused to reheat,
the infants metabolic rate decreased to minimal heatproducing levels, and this decrease occurred even before
the infants cooled body temperature returned to normal.
This was confirmed by Adamsons et al. (21) and again in
the study by Grausz (37).
The study by Adamsons et al. in particular provided
some insight into why homeothermic reactions are not
predicted only by examination of static body temperatures.
In this study, the metabolic rates of infants in various
thermal environments were determined and correlations
computed to determine the relative value of measuring
only rectal temperature, skin temperature, or environmental temperature, or only the difference between the
skin and environmental temperatures, in reliably predicting what the infants metabolic rates actually were at the
time the temperature measurements were made. The
study determined that no correlation existed between rectal temperature and metabolic rate, a slightly better correlation existed between environmental temperature and
metabolic rates, a still better correlation with skin temperature existed, and an almost perfect correlation was
demonstrated between metabolic rate and the difference
between skin and incubator environmental temperatures
(Fig. 4). These results can be understood by recalling that
when body temperature is stable, heat production or metabolic rate must be equal to the rate of heat loss from the
infant. If this balance does not exist, the infant will get
either warmer or cooler, depending on the direction of the
imbalance. So if heat production equals heat loss and heat
loss is proportional only to the difference between the
magnitude of the skin and environmental temperatures
and not to the magnitudes themselves, it follows that heat
production must also be related to the same temperature
difference and, similarly, should be relatively independent
of any single absolute temperature value.
These discoveries led to an approximate definition of
what might constitute an optimal thermal environment in
which to raise small infants. This environment is called a
neutral thermal environment, referring to that set of thermal conditions existing when an infant is in a minimal
metabolic state and has a body temperature that is within a
normal range.
From the previous discussion, it might seem reasonable
that to provide an infant with a neutral thermal environment it is necessary then only to establish skin-to-environment temperature gradients documented in various
published studies to be associated with minimal metabolic
rates in infants with normal temperature. Unfortunately,
such references can provide only very rough guidelines,
since any one infant can achieve different minimal rates of
metabolism at different times and, if body temperature is to
be kept stable, each change in this minimal achievable heat
production rate must be balanced with a change in the
amount of heat loss allowed. When the minimal heat
production increases, the skin environmental temperature
149
Figure 4. Metabolic rate expressed as oxygen consumption (Vo2)

as correlated with rectal temperatures, skin temperature,
incubator environmental temperature, or the difference between
the skin and environmental temperature (DTse). Reproduced by
permission from P. H. Perlstein, Routine and special care
Physical environment. In A. A. Fanaroff and R. J. Martin
(Eds.), Behrmans Neonatal-Perinatal Medicine, 3rd ed., St.
Louis, MO: C. V. Mosby Co., 1983. Adapted from Adamsons
et al. (21).
difference needs to be increased, and when the minimal

heat production falls, the gradient needs to be decreased.
Although concepts such as the neutral thermal environment can be discussed using static equations, it must be
remembered that they actually are used only in dynamic
settings.
In any case, it is very difficult to provide an infant with
truly neutral thermal conditions and becomes practically
impossible in some common situations, such as when head
hoods or other auxiliary gas delivery devices are used
during care. Head hoods are small plastic boxes or tents
made to enclose the infants head when resting on a mattress. The hoods are used to deliver and control the concentration of humidified oxygen to the infant. Since the
head of an infant can represent 20% of the infants body
surface, a significant amount of body heat can be lost if the
head if the head hood temperature is not carefully controlled. Even if the temperature is controlled by prewarming the oxygen prior to its delivery into the head hood, the
infant can lose heat if the gas is delivered at a flow rate that
produces an excessive wind chill component to the convective heat flux. It also has been documented that even when
total body heat losses are less than usually needed to
stimulate a homeothermic response, any local cooling of
facial skin tissue can stimulate a baby to become hypermetabolic (36,38).
150
INCUBATORS, INFANTS
Since no studies have been published to provide guidelines for establishing neutral thermal conditions in an
incubator containing auxiliary sources of heating and cooling, and because such sources are very commonly used
during infant care, no incubator manufacturer can guarantee, when an auxiliary devices are used, that such conditions can be produced by any incubator on the market
today. As a corollary, unless both body temperatures and
infant metabolic rates are continuously monitored, infant
caretakers and medical researchers are similarly constrained from claiming precision in their delivery of continuous and certifiable thermal protection that is
consistent with the concept of thermoneutrality.
Before we leave this subject, it should also be noted that
a significant number of knowledgeable baby care specialists disagree with the idea that a neutral thermal environment represents an optimal goal for incubator control.
Their arguments are numerous, but most often include the
irrefutable fact that no one has ever documented scientifically that such protection is really beneficial. They also
cite the studies by Glass et al. (2,3) in which it was
documented that babies tend to lose their very important
self-protective ability to react as homeotherms if not
exposed to periodic cold stresses. This means that one price
paid by an infant raised in a neutral thermal environment
is adaptation to that environment and, much as a prolonged stay in the tropics diminishes an adults capacity to
tolerate the northern winters, a baby so adapted may be
more susceptible to the damaging effects of unavoidable
occasional exposures to the cold.
It must be emphasized that the arguments against the
neutral thermal environment are not arguments in favor of
letting all babies stay cold; the debate primarily concerns
whether a baby is better off in an environment that
theoretically maximizes growth by reducing metabolic
work to an absolute minimum but might increase the
infants susceptibility to subsequent stresses, or better
off in an environment that very mildly stimulates the
infant to metabolically contribute to his own ongoing thermal welfare so that important self-protective capabilities
are not forgotten. As yet there are insufficient scientific
data to resolve this issue. A more recent observation is that
the body temperatures of the fetus and older infant are
higher than the typical neutral thermal environment proposed for preterm infants, and in both cases, follow a
circadian rhythm that is not observed or supported in
the typical infant incubator. These considerations imply
that while the neutral thermal environment is a useful
concept for current incubator design, it is not yet known
how the optimal thermal environment should be defined,
especially for the preterm infant.
INCUBATOR STUDIES
In spite of the difficulties encountered when attempting to
define, let alone achieve, the optimal environmental conditions that are protective for small babies, it is clear that
babies raised in different environments do have different
survival rates. The scientific studies documenting these
differences in survival in different environments are worth
reviewing, since they have provided insight into features

distinguishing some incubators from others and ways in
which these features may produce environments that can
prove to be both protective to some infants and dangerous
for others. These studies have also been the major impetus
to changes in designs that have resulted in the kinds of
incubator devices in use today.
With few exceptions, until the early 1970s, incubator
designers relied only on convective heaters to warm the
chamber within which a baby was contained. Such devices
were relatively simple to construct and provided a method
whereby the chamber mattress could be kept warm thereby
limiting conductive heat losses, and a method to keep the
surrounding air warm, limiting convective losses. The use
of wet sponges early in the history, and later evaporation
pans, over which the convective currents of warmed air
passed before entering the infants chamber, provided the
humidity needed to limit evaporative losses. Additionally,
the warmed air contained in the chamber produced some
warming of the chamber walls thereby reducing to some
degree radiant heat losses from the infant. The heating
units in early models of these incubators were controlled
only by simple air temperature-sensitive thermostat
mechanisms.
Such an incubator with clear plastic walls for enhancing
visualization of the contained infant was used in a famous
series of infant survival studies published between 1957
and 1965. In this incubator, a fan was used to force the
convective air currents into the infants chamber after
passage over a heating element through a turbulence
producing baffle resting in a humidifying pan of water.
A highlight of these studies was published in 1958 by
Silverman et al. (7) who compared the survival rates of
two groups of premature infants cared for in this convectively heated and humidified device. For one group of
infants the incubator air was heated to 28 8C and for the
other group the air was heated to a warmer 32 8C. The
study resulted in a conclusion that infants cared for in the
warmer incubator had a higher survival rate than did
infants cared for in the cooler incubator. During the study
it was observed that although babies in the warmer incubator had a greater chance of surviving, not all of the
infants who survived in the warmer incubator had warm
body temperatures; in fact, 10% of the babies in the 32 8C
incubator had body temperatures 35.5 8C. Dr. Silverman
deduced that the reason some of the babies got cold was due
to excessive radiant heat losses to the thin plastic chamber
walls that were cooled by virtue of their exterior surfaces
being exposed to the cooler nursery environment.
Dr. Silverman wished to test whether even better survival rates could be achieved by assuring that an infant
was not only in a warm incubator, but that the infants body
temperature also was always kept within a normal range.
Along with Dr. Agate and an incubator manufacturer he
helped develop a new incubator that was radiantly heated
to reduce the radiant losses observed when the incubator
was only convectively heated (39). To assure that the
contained infants temperature was maintained within
normal range, the new incubator employed an electronic
feedback servo-control mechanism that responded to
changes in the temperature of a thermistor attached to
INCUBATORS, INFANTS
151
Figure 5. Logic leading to development of skin servo-controlled radiantly heated convectively

ventilated incubator. (1) An unprotected baby loses heat from skin surfaces by conduction,
convection, evaporation, and radiation. (2) A radiant heater eliminates radiant and conductive
losses, but not those caused by convection and evaporation. (3) An unhumidified convectively heated
incubator eliminates convective and conductive losses, but not those caused by radiation and
evaporation. (4) Humidifying a convectively heated incubator eliminates all major losses except
for the losses by radiation. (5) Using a radiant heater to warm a convectively ventilated and
humidified incubator should eliminate all sources of heat loss from the infants skin. (6) Normal
infant temperature can be ensured by adding a controller to the incubator so that power is delivered
to the radiant heater whenever the infants skin temperature falls below a preset value. Reproduced
by permission from P. H. Perlstein, Routine and special carePhysical environment. In A. A.
Fanaroff and R. J. Martin (Eds.), Behrmans Neonatal-Perinatal Medicine, 3rd ed., St. Louis, MO: C.
V. Mosby Co., 1983.
the infants skin surface, causing the incubators radiant

heater to turn on or off, depending on whether the transduced skin temperature value was below or above an
absolute temperature value considered normal (Fig. 5).
Note that before settling on a servo-controlled, radiantly
heated and convectively ventilated system Agate and Silverman did explore alternative methods whereby an incubator could be equipped to guarantee that an infants
temperature was maintained within a normal range. In
particular, they considered simply using the wellestablished convective heating system in the servo-control
loop but rejected this approach when they discovered that
when servo controlled in response to changes in skin
temperature, the convective heater produced massive
and unacceptable changes in air temperature within the
incubator chamber. The servo-controlled radiant heater,
however, produced an environment in which the air temperature was quite stable, especially when compared to the
thermal cycling recorded within the convectively heated
servo-controlled system.
The radiantly heated, convectively ventilated, and skin

servo-controlled enclosed incubator was evaluated in two
independent studies, with results published in 1964 (9,10).
In these controlled trials, premature infants were divided
into two groups: one group of infants was provided care in
the new radiantly heated incubator that was servo
controlled to maintain the contained infants skin temperature at 36 8C, while the other group of like babies was cared
for using the simpler 32 8C air temperature thermostatcontrolled convectively heated incubator that Silvermans
group concluded was the best incubator setup in their
previous study published in 1958. The two studies in
1964 reached a common conclusion; the skin temperature-controlled radiantly heated system produced higher
survival rates when used during the care of the infants
studied. Because of fabricating difficulties, though, this
radiantly heated incubator model was commercially marketed for only a short period of time; soon after its introduction, it was replaced on the commercial market
by a skin servo-controlled convectively heated enclosed
152
INCUBATORS, INFANTS
incubator that was easier to fabricate and, like the

radiantly heated device, was also capable of keeping an
infants skin temperature at a value considered normal.
The introduction of this convectively heated servocontrolled device on the market was justified by an extrapolated interpretation of the studies reported in 1964. This
common interpretation led to a conclusion that the studies
simply demonstrated that, in terms of survival, it was only
important to keep a babys skin temperature warm and
stable. The interpretation ignored the fact that more than
just a difference in skin temperatures distinguished the
two study groups. Infants in the two different study environments, the one produced by an air temperature referenced thermostat controlling a convective heater and the
other by a skin temperature referenced servo system controlling a radiant heater, were also, as previously well
discussed by Agate and Silverman, exposed to environments that differed in the frequency and amplitude of
thermal cycling produced by the different systems (39).
The radiantly protected infants, who survived better, not
only had warmer and more stable skin temperatures as a

group, but were also exposed to a much more stable environment than were the convectively protected infants with
the less favorable group outcomes.
The commercially released convectively heated and skin
temperature referenced servo-controlled incubator was the
most common incubator in clinical use during the late
1960s; not until 1970 was the characteristic rapidly changing air temperatures within the incubator chamber redescribed and shown to cause some sick small babies to stop
breathing (40). These episodes of respiratory arrest, called
apneic spells, were specifically observed during the incubators heating cycles. The mechanism whereby a sudden
rise in temperature causes some babies to become apneic
remains unknown, but was an observation well reported
even prior to the 1970 publication. Even without knowing
the mechanism of this relationship, it remains undisputed,
so incubator manufacturers have continued to search for
ways to produce stabilization of the incubator environmental temperatures (Fig. 6).
Figure 6. Skin and air temperature characteristics recorded using four different incubator
systems.(1) A convectively heated and humidified incubator in which air temperature is
thermostatically controlled. This was the device studied by Silverman in 1958 (7). Note cyclie
variations in air temperature and wide variation in recorded skin temperatures.(2) A radiantly
heated convectively ventilated and humidified incubator that is servo controlled to maintain skin
temperature at specified value. This was the device studied by Day (9) and Beutow (10) in 1964. Note
that the walls are warm, limiting radiant heat losses and that air temperature is stable and skin
temperature variations are minimal. (3) A convectively heated and humidified incubator in which
the air temperature heating is servo controlled to maintain skin temperature at specified value. This
was the device reported to cause some babies to stop breathing (40) as a response to erratic heating
cycles that produces rapid increases in air temperature. (4) A convectively heated and humidified
incubator that is computer controlled using Alcyon algorithm (11). Note the stable air temperature
and minimal variability in skin temperature.
INCUBATORS, INFANTS
INCUBATOR DYNAMICS
There are many reasons why attempts to stabilize incubator heating have been only partially successful. It is
fairly simple to build a box within which the temperatures
can be predictably controlled and stabilized for prolonged
periods of time if the box is never opened and the thermal
characteristics of both the box and its contents never
change, but these simplifying conditions never exist when
caring for a sick infant. When infants are cleaned, fed,
examined, or otherwise cared for, they must be touched,
and the incubator box must be entered. Infants body
positions frequently change, exposing different surface
areas with different shapes to the incubator environment
causing changes in convective flow patterns and altering
the view factors influencing radiant heat flow to the incubator walls. Incubator openings necessitated by the need to
touch a contained infant cause both environmental cooling
and increased infant heat loss that, in an incubator heated
in response to either air temperature or infant temperature
changes, causes a logical increase in the incubators heat
output. If any such heating requirement is sustained, the
incubator heating element warms to the point where it will
retain and release heat even after the incubator is closed
and the need for additional heating has passed. Such heat
retention and release contributes to what is commonly
referred to as the thermal lag characteristic of a heating
system and can cause temperatures to overshoot, that is to
rise above the temperature level targeted when the heater
was activated. This is the same phenomenon observed
when an electric stove is turned off, and yet the heating
element continues to glow brightly prior to cooling. As with
an electric stove heating element, the heater in an incubator is not only slow to cool, but also relatively slow to
warm up when first energized after the heater is turned on.
Again, just as unintended overheating can occur, the thermal lag due to the mass of the heater can result in an
incubator getting colder than intended because of this
characteristic delay between action and reaction.
Besides the heater element, there are numerous other
places where heat is stored in the incubator system. For
example, heat storage occurs in the water used for humidification and in the air masses between the heater and the
incubator chamber. Since these must be heated to a temperature higher than the incubator chamber in order to
raise the chamber temperature, the heat stored in these
parts also will continue to raise the chamber temperatures
even after the heater power supply has been turned off.
Conversely, when the heater power is turned back on, not
only the heater but also the air in the spaces leading to the
chamber must heat up before the temperature of the air in
the chamber can rise.
It is these delays between the time the heater power is
changed and the time the air temperature responds that
determines the magnitude and frequency of the air temperature cycles to which an incubated infant is exposed.
Thermal lag obviously contributes to the tendency for
incubator environments to become unstable and, thereby,
potentially threatening to the contained infant. Many
hardware and logical software solutions to this problem
have been tried in commercially available devices, but all
153
have been frustrated by the complex nature of newborn

care, which results in a degree of unpredictability beyond
the compensating capability of any solution thus far tried.
The implementation of feedback control on incubator heating is an example of one way to attempt to respond to many
of these problems. However, examining how servo mechanisms actually work in a little more detail provides a deeper
appreciation of why this logical technology often fails in the
dynamic setting of an incubator and may even contribute to
instability in the incubator environment.
Feedback Control
Feedback control systems are commonly referred to as
closed loop control systems, as contrasted to open loop
systems. A cooking stove again can be used to give an
example of each type of control system. Stove top heaters
are typically controlled by an open loop system. That is, a
dial is adjusted controlling the quantity of gas or electricity
going to the heater unit. In this manner, a fixed rate of heat
production by the heater unit is specified. This is called an
open-loop control system because after once setting the
rate of heat production, the heater will continue to produce
the same heat output regardless of how hot the object on
the stove gets.
In contrast, the oven of a modern stove is equipped with
a closed-loop temperature control system, in which a dial is
adjusted to specify the desired oven temperature. In control system parlance, this specified temperature is referred
to as the set point. A temperature sensor inside the oven
works in conjunction with the temperature setting dial to
control the rate of heat production in the oven heating unit
and when the temperature measured by the oven temperature sensor rises to the set point value on the control dial,
the oven heat production is reduced or stopped entirely.
After the heat production is stopped, the oven temperature
slowly falls as the heat escaped from the oven to the
surrounding area. At some point, the oven temperature
will fall below the temperature set on the control and the
heater will again be turned on; this onoff cycling will
continue as long as the oven is in operation.
Feedback Control and Incubators
An incubator heated in automatic feedback response to
changes in electronically transduced infant temperature is
called an infant skin servo-controlled (ISC) incubator. In
one type of ISC incubator the heater is instructed to turn
completely on when the infants skin temperature falls
below a preset lower limit or turn completely off when skin
temperature exceeds a defined upper limit. Because power
applied to the heater is either maximal or zero, this form of
servo mechanism is called nonlinear or onoff control.
Another form of control is designated as linear proportional
servo control. In a proportional control system the amount
of power applied to the incubator heater is graduated in a
manner to be proportional in some linear fashion to the
transduced skin temperature deviation from a predetermined value. The amount of power actually applied to the
heater for each incremental change in skin temperature
can be different in different realizations of this control
method, and this increment determines the amount of
154
INCUBATORS, INFANTS
deviation in skin temperature that can occur before full or

zero power is applied.
The theoretical advantage of a proportional over an
onoff servo-control system is the possibility of limiting
the tendency that a large heating element has to overshoot
or undershoot a desired heater temperature. In a proportional system, the heater is turned off slowly as the infants
skin temperature warms toward the set point temperature.
This contrasts with an onoff system that remains fully
energized until the set point is reached. In an incubator
system, a properly designed skin temperature referenced
proportional control unit will produce very stable environmental temperatures as long as the infant is stable, the
temperature sensing the thermistor attached to the skin
remains undisturbed, and the incubator chamber is kept
closed and protected from outside environmental perturbations. In a clinical setting such stable and undisturbed
conditions exists infrequently, so the theoretical advantages of proportional control over onoff control are difficult to demonstrate. In fact, the cycling of temperatures
recorded within a proportionally controlled incubator in a
dynamic clinical setting can be indistinguishable from that
recorded in an onoff controlled incubator, and this functional behavior is predicted in basic feedback control theory
(Fig, 7).
The thermal cycles recorded in servo-controlled incubators are produced as a combined consequence of (1) an
inherent characteristic of closed-loop feedback systems; (2)
periodic extreme perturbations of the skin thermistor that
trigger either full or zero energization of the incubator
heater; and (3) the incubators thermal lag characteristic,
which causes repetitive over-and undershoot of temperatures targeted by the control logic. Following the initiation
of a cycling pattern, the time it takes the system to settle
down and reestablish stable control is variable and related
to the heat-dissipating speed of the heater material and the
thermal buffering capability of the homeothermic infant.
In some situations, the cycling, when induced by a single
perturbation, has been observed to continue for many
Figure 7. Example of variations in dynamic air temperature

changes when an enclosed incubator that was initially servo
controlled to maintain a stable air temperature was switched to
a skin temperature referenced servo control mode and then
was perturbated by changing the site of attachment of the
skin temperature-sensing thermistor. The wide thermal cycling
illustrated as a consequence of this sequence continued for 3 h
before returning to a more stable pattern.
hours and, when an incubator is repeatedly perturbated,

for many days. The published evidence that some babies
react to these thermal cycles by becoming apneic justifies a
reminder that these characteristic thermal cycles represent a profound problem negating some of the advantages
intended when feedback control is applied in incubator
designs. At least in incubator servo systems that use skin
temperature as a reference value to determine heater
status, even the theoretical negative effects often outweigh
any theoretical or demonstrable effects that are positive.
This can be appreciated by recalling that a sine qua non of
optimal control in a skin temperature referenced servo
system is the reliable and accurate measurement of an
infants skin temperature and, using todays technology in
a clinical setting, the ability to transduce skin temperatures accurately for prolonged periods of time is marginal
at best and, perhaps, even impossible.
Skin Temperature Measurement
Skin temperature measurement accuracy is limited by
variability in the characteristics of infant, transducers,
and infant care techniques. The surface temperatures of
infants are not homogeneous because of difference in (1)
skin and subcutaneous tissue thickness over different body
parts, (2) differences in structures underlying different
skin surfaces, and (3) difference in the vascularity and
vasoreactivity-characterizing different body regions.
Different skin surfaces have different temperatures, and
when temperature transducers are connected to the skin
surface, they measure the temperature only at the specific
site of their connection. Moreover, thermistors are
attached using devices that can compression of superficial
skin vessels underlying the thermistor element. Thus, by
their very attachment, thermistors modify both the absolute temperature they are expected to measure and the
spontaneous dynamic variability in the measured skin
temperature that is normally affected by the changing
amounts of warm blood flowing through the skin over
different time periods. Additional factors also affect thermistor accuracy. Thermistors are faulted as precise
skin temperature measuring devices because they are
manufactured in various shapes and sizes and are protected using different materials so that each affects transduction in a specific and different way. Thermistors also
generally measure temperature three dimensionally (3D).
They are affected not only by the temperature of the surface to which they are attached, but also by the environment to which their unattached surfaces are exposed.
Depending on the amount and type of insulation used in
their manufacture, they are also affected by the temperature of the wires used to connect the thermistor to electronic signal conditioners.
These inherent characteristics of thermistors, added to
the fact that they are freely moved during clinical care from
one site of attachment to another, provide sufficient cause
to explain why the skin temperature-dependent heatercontrolling servo mechanism in an incubator can easily be
directed into an unstable state that produces thermal
cycling in the environment. This environmental instability
is even further exacerbated by infant care practices that,
INCUBATORS, INFANTS
for examples, cause thermistors to be removed from the

skin when X rays are taken, or to be covered with sterile
towels during surgical procedures. During such carerelated events, the thermistor is fooled into measuring
environmental and not skin temperature. Obviously, if
the thermistor provides the servo electronics with misinformation, the servo control decisions based on this information can be nonsensical.
THE NONTHERMAL ENVIRONMENT
Although infant incubators were initially designed solely to
maintain body temperature in high risk infants, they are
now seen in a much more complex role, as devices that
provide a complete microenvironment for high risk
infants. To one extent or another, they are used to modify
the sensory input an infant receives through visual, auditory, olfactory, and kinesthetic pathways. In addition,
incubators are now appreciated as the source of exposure
to potentially unwanted environmental toxins, such as
electromagnetic radiation (EMR) and chemical compounds
used in the manufacture and operation of the incubator.
Closed incubators are often used as delivery systems for
oxygen and humidification, sometimes in ways unanticipated at the time of their design and manufacture.
Incubators have been developed for specialized purposes,
such as transport, use in an magnetic resonance imaging
(MRI) suite, or cobedding twin infants. Incubators have
increasingly been designed as a platform for a modular
system of support equipment including ventilators, IV
pumps, and monitors. As these devices become increasingly controlled by digital components, they also gain the
capability of integrating their data output, so that incubator-derived information, such as air temperature and the
infants skin temperature, can be continuously recorded in
an electronic medical record. Incubators have had a role in
infection control since their earliest days, but this function
is now being increasingly emphasized as infection has
emerged as the leading cause of late morbidity in premature infants. These multiple functions of modern incubators increase their complexity exponentially, since many of
these factors interact with one another, often in ways
unanticipated by either the designers or the users. Because
of the recent nature of these nonthermal applications of the
infant incubator, there is only a limited scientific foundation to guide designers and caregivers, so not all of these
topics will be discussed in greater depth below.
The Infant Incubator as a Sensory Microenvironment
Although the comparison of an incubator to a uterus, the
environment it replaces, has been noted since the earliest
days of incubator design, it will be evident from the preceding sections of this article that temperature regulation
has been the first and primary design consideration: and
necessarily so, since it had immediate implications for
infant survival. When incubators became used as devices
for delivery of supplemental oxygen in the mid-twenteth
century, morbidity and mortality were again the primary
endpoints: first for improved survival as the benefits of
oxygen supplementation were identified, and then for
155
increased morbidity as an epidemic of oxygen-induced

blindness from retinopathy of prematurity followed, again
chronicled most eloquently by Dr. Silverman (41). Only
recently have the other environmental features of the
uterus been compared to the micro and macro environments of the NICU and the implications for design been
considered.
Taste, smell, and touch are the earliest fetal senses to
develop, beginning in the second trimester of pregnancy,
followed closely by auditory development, and finally by
visual development as the baby approaches term gestation.
While thus far there appears to be no reason to suspect that
the sense of taste is stimulated or influenced by the incubator, there is accumulating evidence that the other senses
are indeed affected by this microenvironment.
Infants can develop a conditioned response to certain
odors that they come to associate with painful procedures,
such as alcohol, whereas a pleasant odor has been shown to
reduce apnea, and babies will orient preferentially to odors
from their mother (42).
In utero, infants are exposed to a fluid environment,
frequent movement with circadian rhythmicity, and as
they approach term, increasing contact with a firm boundary, features absent in the typical incubator. After-market
adaptations that caused the bed or the entire incubator to
move in one fashion or another have been introduced
sporadically, often with the intent of reducing infant
apnea, but none have been documented to be efficacious.
Additional modifications of the infant mattress to make it
more suitable for the skin and developmental needs of
preterm infants have been introduced, again without clear
benefit to this point.
Sound is a constant although variable stimulus in the
uterus, and different in many ways from that of the modern
incubator. In utero sounds are delivered via a fluid medium
where lower pitched sounds predominate, and the mothers
voice, heartbeat and bowel sounds are far more prevalent
than any extraneous noise. As such, there is a definite
circadian rhythm both to the sounds and to movement
associated with them. In the closed incubator, the predominant sound is that of the incubator fan that produces a
constant white noise, usually in excess of 50 dB, a level that
has been shown to interfere with infant sleep (43) and
speech recognition (44). Depending on the NICU in which
it is used, this may be louder or softer than the NICU
ambient noise level, so the closed incubator may be used
as a haven from a noisy NICU, or itself could be noisier than
the external environment, in which case the open radiant
warmer might provide a more suitable auditory environment. A number of after-market modifications have been
suggested to reduce both fan noise and intrusion of noise
from outside the incubator, including blankets or quilts to
cover the incubator and sound-absorbing panels (45,46), but
their use is problematic to the degree that they affect air flow
and other operating characteristics of the incubator.
The visual environment of the incubator may be presumed to be neutral, but incubators are often used to
control the visual environment of the NICU, particularly
in conjunction with the use of incubator covers, to produce
a dimly lit environment which may enhance infant sleep
(47). Light penetration into the incubator may also affect
156
INCUBATORS, INFANTS
circadian rhythmicity in the preterm infant (48), and may

be a source of heat gain through the greenhouse effect.
Electromagnetic Radiation in the Infant Incubator
Any electrically powered device emits electromagnetic
radiation (EMR), usually at intensities far below those
considered to constitute a risk. Since the organs of preterm
infants are in a crucial and rapid phase of growth, however,
concern about EMR emitted by incubator and radiant
warmer heaters and other components has merited special
attention. Several studies have documented EMR levels in
incubators, with proposed strategies to reduce exposure
including shielding panels (49) and increasing the distance
between the baby and the EMR source (50).
Incubators for Specialized Purposes
The conventional closed incubator or radiant warmer is
used as a static device in the NICU, but the same needs for
temperature control and a safe microenvironment exist
for infants who require transport from one hospital to
another, or to an MRI suite. Transport incubators place a
premium on space (so that multiple modular components
can be mounted) and weight (especially those used for air
transport). Incubators developed for use in an MRI suite
must be similarly portable, but use almost exclusively
plastic materials and have an integrated coil for scanning
(51,52).
SUMMARY
Infant incubators are specially heated devices that provide
a bed surface or chamber within which an infant can be
cared for and kept warm. Throughout this article both the
positive and negative features of todays incubators have
been noted and placed into both a context of what is
theoretically desirable and of what is practically feasible.
It is apparent that, at present, our knowledge of infant
physiology and the availability of technical solutions are
severely limited, and that all existing incubator devices
reflect and are faulted by these limitations. In historical
perspective, however, it is clear that incubators over the
past 100 years have steadily been improved by manufacturers to incorporate new items of knowledge and
technology as they become available. This historical
path has been fruitful and provides, in its continuing
course, direction for the future in incubator development.
Future iterations of the infant incubator will need to
incorporate new information on the optimal microenvironment for the high-risk newborn as well as new capabilities made possible by the ongoing quantum changes in
digital technology.
BIBLIOGRAPHY
1. Meysta n J, Ja rai I, Fekete M. The total energy expenditure
and its components in premature infants maintained under
different nursery and environmental conditions. Pediatr Res
1968;2:161.
2. Glass L, Silverman WA, Sinclair JC. Effect of the thermal

environment on cold resistance and growth of small infants
after the first week of life. Pediatrics 1968;41:1033.
3. Glass L, Silverman WA, Sinclair JC. Relationship of thermal
environment and caloric intake to growth and resting metabolism in the late neonatal period. Biol Neonate 1969;14:324.
4. Gandy GM, et al. Thermal environment and acid base homeostasis in human infants during the first few hours of life. J
Clin Invest 1964;43: 751.
5. Cornblath J, Schwartz R. Disorders of carbohydrate metabolism in infancy. In: Shaffer JA, editors. Major Problems in
Clinical Pediatrics. Vol. 3. Philadelphia: Saunders; 1966. p 34.
6. Silverman WA, Blanc WA. The effect of humidity on survival of
newly born premature infants. Pediatrics 1957;20:477.
7. Siverman WA, Ferrig JW, Berger AP. The influences of the
thermal environment upon the survival of newly born premature infants. Pediatrics 1958;22:876.
8. Silverman WA, Agate FJ, Ferrig JW. A sequential trial of the
nonthermal effect of atmospheric humidity on survival of
newborn infants of low birth weight. Pediatrics 1964;34:171.
9. Day RL, Caliguiri L, Kamenski C, Ehrlich F. Body temperature and survival of premature infants. Pediatrics 1964;34:
171.
10. Beutow KC, Klein SW. Effect of maintenance of normal skin
temperature on survival of infants of low birth weight. Pediatrics 1964;34:163.
11. Perlstein PH, Edwards NK, Atherton HD, Sutherland JM.
Computer assisted newborn intensive care. Pediatrics 1976;
57:494.
12. Evaluation of infant radiant warmers. Health Devices,
1973;3:4.
13. Perstein PH. Thermal control. Rep Ross Conf Pediatr Res
1976;69:75.
14. Edwards NK. Radiant warmers. Rep Ross Conf Pediatr Res,
1976;69:79.
15. Evaluation of infant incubators. Health Devices. 1981;11:47.
16. Evaluation of infant incubators. Health Devices 1982;11:191.
17. Wu PYR, Hodgman JE. Insensible water loss in preterm
infants. Pediatrics 1974;54:704.
18. Cone Jr. TE, History of the Care and Feeding of the Premature
Infant. Boston: Little, Brown; 1985.
19. Scopes JW. Metabolic rate and temperature control in the
human body. Br Med Bull 1966;22:88.
20. Scopes JW, Ahmed I. Minimal rates of oxygen consumption in
sick and premature newborn infants. Arch Dis Child
1966;41:407.
21. Adamsons K Jr., Gandy GM, James LS. The influence of
thermal factors upon oxygen consumption of newborn infants.
J Pediatr 1965;66:495.
22. Hill JR, Rahimtulla KA. Heat balance and the metabolic rate
of newborn babies in relation to environmental temperature:
And the effect of age and weight on basal metabolic rate. J
Physiol(London) 1965;180:239.
23. Dawes GS. Oxygen consumption and temperature regulation
in the newborn. I Foetal and Neonatal Physiology. Chicago:
Year Book Medical Publishersh; 1968. p 191.
24. Cross K, et al. Lack of temperature control in infants with
abnormalities of the central nervous system. Arch Dis Child
1971;46:437.
25. Dawkins MJR, Hull D. Brown fat and the response of the
newborn rabbit to cold. J Physiol (London) 1963;169:101.
26. Hill JR. Oxygen consumption of newborn and adult mammals:
Its dependence on oxygen tension in inspired air and on environmental temperatures. J Physiol (London) 1959;149: 346.
27. Cree JE, Meyer J, Hailey DM. Diazepam in Labour: Its
metabolism and effect on the clinical condition and thermogenesis of the newborn. Br Med J 1973;3:251.

28. Bruck K. Heat production and temperature regulation. In:
Stave U, editor. Pernatal Physiology. New York: Plenum
Press; 1978. p 474.
29. Day RL. Respiratory metabolism in infancy and childhood. Am
J Child 1943;65:376.
30. Hey EN, Katz G. Evaporative water loss in the newborn baby.
J Physiol (London) 1969;200:605.
31. Sulyok E, Jequier E, Ryser G. Effect of relative humidity on
thermal balance of the newborn infant. Biol Neonate 1972;
21:210.
32. Belgaumkar TR, Scott KE. Effects of low humidity on small
premature infants in servo control incubators. Biol Neonate
1975;26:337.
berg PA, Sedin G. Transepi33. Hammarlund K, Nilsson GE, O
dermal water loss in newborn infants. Acta Paediatr Scand
1977;66:553.
berg PA, Sedin G. Transepi34. Hammarlund K, Nilsson GE, O
dermal water loss in newborn infants: Evaporation from the
skin and heat exchange during the first hours of life. Acta
Paediatr Scan 1980;69:385.
35. Hey EN, Mount LE. Heat losses from babies in incubators.
Arch Dis Child 1967;42:75.
36. Bruck K. Temperature regulation in the newborn infant. Biol
Neonate 1961;3:65.
37. Grausz JP. The effects of environmental temperature changes
on the metabolic rate of newborn babies. Acta Paediatr. Scand
1968;57:98.
38. Mestan J, Jrai I, Bata G, Fekete M. The significance of facial
skin temperature in the chemical heat regulation of premature
infants. Biol Neonate 1964;7:243.
39. Agate FJ, Silverman WA. The control of body temperature in
the small newborn infant by low-energy infra-red radiation.
Pediatrics 1963;37:725.
40. Perlstein PH, Edwards NK, Sutherland JM. Apnea in premature infants and incubator air temperature changes. N
Engl J Med 1970;282:461.
41. Silverman WA. The lesson of retrolental fibroplasias. Sci Am
1977;236:100.
42. Schaal B, Hummel T, Soussignan R. Olfaction in the fetal and
premature infant: Functional status and clinical implications.
Clin Perinatol. 2004;31:261.
43. Morris BH, Philbin MK, Bose C. Physiologic effects of sound on
the newborn. J Perinatol 2000;20:S55.
44. Robertson A, Stuart A, Walker L. Transmission loss of sound
into incubators: implications for voice perception by infants. J
Perinatol 2001;21:236.
45. Johnson AN. Neonatal response to control of noise inside the
incubator. Pediatr Nurse 2001;27:600.
46. Bellini CV, et al., Use of sound-absorbing panel to reduce noisy
incubator reverberating effects. Biol Neonate 2003;84:293.
47. Hellstrom-Westas L, et al., Short-term effects of incubator
covers on quiet sleep in stable premature infants. Acta Paediatr 2001;90:1004.
48. Rivkees SA. Emergence and influences of circadian rhythmicity in infants. Clin Perinatol 2004;31:217.
49. Bellieni CV, et al. Reduction of exposure of newborns and
caregivers to very high electromagnetic fields produced by
incubators. Med Phys 2005;32:149.
50. Bellieni CV, et al. Increasing the engine-mattress distance in
neonatal incubators: A way to decrease exposure of infants to
electromagnetic fields. Ital J Pediatr 2005;29:74.
51. Blumi S, et al. MR imaging of newborns by using an MRcompatible incubator with integrated radiofrequency coils:
Initial experience. Radiology, 2004;231:594.
52. Whitby EH, et al. Ultrafast magnetic resonance imaging of the
neonate in a magnetic resonance-compatible incubator with a
built-in coil. Pediatrics 2004;113:e150.
157
Further Reading
Adamsons K. The role of thermal factors in fetal and neonatal life.
Pediatr Clin North Am 1966;13:599.
Ahlgren EW. Environmental control of the neonate receiving
intensive care. Int Anesthesiol Clin 1974;12:173.
Bruck K. Heat production and temperature regulation. In: Stave
U, editor. Perinatal Physiology New York: Plenum Press; 1978
p 455.
Dawes GS, Oxygen consumption and temperature regulation in
the newborn. I Foetal and Neonatal Physiology. Chicago: Year
Book Medical Publisher; 1968. p 191.
Delue NA. Climate and environment concepts. Clin Perinatal
1976;3:425.
Hey EN, Katz G. The optimum thermal environment for naked
babies. Arch Dis Child 1970;45:328.
Holman JP. Heat Transfer, New York: McGraw-Hill; 1981.
Klaus M, Fanaroff A, Martin RJ. The physical environment. In:
Klaus MH, Fanaroff AA, editors. Care of the High Risk Neonate. Philadelphia: Saunders; 1979. p 94.
Lutz L, Perlstein PH. Temperature control in newborn babies.
Nurs Clin North Am 1971;6:15.
Mayr O. The Origins of Feedback Control. Cambridge, (MA): MIT
Press; 1970.
Ogata K. Modern Control Engineering, Englewood Cliffs (NJ):
Prentice-Hall; 1970.
Oliver TK. Temperature regulation and heat production in the
newborn. Pediatr Clin North Am 1965;12:765.
Oppenheim AV, Willsky A, Young IT. Signals and Systems, Englewood Cliffs (NJ): Prentice-Hall; 1983.
Perstein PH. Thermal regulation. In: Fanaroff AA, Martin RJ,
editors. Behrmans Neonatal-Perinatal Medicine, 3rd ed. St.
Louis (MO): Mosby; 1983. p 259277.
Scopes JW. Thermoregulation in the newborn. In: Avery GB,
editors. Neonatology, Philadelphia: Lippincott; 1975. p 99.
Sinclair JC. The effect of the thermal environment on neonatal
mortality and morbidity. In: Adamson K, Fox HA, editors.
Preventability of Perinatal Injury. New York: Alan R. Liss;
1975. p 147.
Sinclair JC. Metabolic rate and temperature control. In: Smith CA,
Nelson NM, editors. The Physiology of the Newborn Infant.
Springfield (IL) : Thomas; 1976. p 354.
Todd JP, Ellis HB. Applied Heat Transfer. New York: Harper &
Row; 1982.
See also BIOHEAT
TRANSFER; NEONATAL MONITORING; TEMPERATURE
MONITORING.
INFANT INCUBATORS.
See INCUBATORS,
INFANT
INFORMATION SYSTEMS FOR

RADIOLOGY. See RADIOLOGY INFORMATION SYSTEMS.
INFUSION PUMPS. See DRUG INFUSION SYSTEMS.
TATSUO TOGAWA
Waseda University
Saitama, Japan
INTRODUCTION
Temperature can affect the electronic characteristics of
semiconductor devices. Although this is a disadvantage
158
in many applications, especially for analogue devices, it

may be turned into an advantage if such a device is used as
a temperature sensor. In principle, any parameter in such
a device having a temperature coefficient can be used for
temperature measurement. For example, a temperature
telemetry capsule, in which a blocking oscillator frequency
varies with temperature, has been developed for measuring gastrointestinal temperature (1). In this system, the
temperature affects the reverse-bias base-collector current, which determines the period of relaxation oscillation.
However, it has been shown that the voltage across a pn
junction of a diode or transistor under a constant forwardbias current shows excellent linear temperature dependency over a wide temperature range. Many conventional
or specially designed diodes or transistors composed of
Ge, Si, or GaAs have been studied for use as thermometers
(24).
The advantages of diodes and transistors as temperature sensors are their high sensitivity and low nonlinearity. The temperature sensitivity under normal operation is
ca 2 mV/K, which is 50 times higher than that of a
copper-constantan thermocouple. The nonlinearity is low
enough for many applications, although its value depends
on the structure and material of the device. It is known that
a Schottky diode, which has a structure composed of a
rectifying metal-semiconductor contact, possesses good
voltagetemperature linearity (5). Some transistors used
as two-terminal devices by connecting the base to the
collector also possess good linearity (6,7), and a transistor
that has been especially developed for temperature sensing
is commercially available (8). This has a linearity that is
comparable to that of a platinum-resistance temperature
sensor.
It is advantageous to have a diode and a transistor
temperature sensor fabricated on a chip with associated
interfacing electronics using integrated circuit (IC) technology. Several integrated temperature sensors that provide either analogue or digital outputs have been developed
and are commercially available. A diode or transistor
temperature sensor fabricated on a central processing unit
(CPU) chip is especially useful when used to monitor the
temperature of the chip. Such a sensor has been used to
detect overheating, and to protect the CPU system by
controlling a fan used to cool the chip or to slow down
the clock frequency.
THEORY
The characteristics of pn junctions are well known (9,10).
In pn junction diodes, the current flowing through the
forward-biased junction is given by
I Is eqV=mkT 1
where Is is the saturation current, q is the electron charge,

V is the voltage across the junction, k is the Boltzmann
constant, m is the ideality factor having a value between 1
and 2, which is related to the dominant current component
under the operating conditions used, and T is the absolute
temperature. At a temperature close to room temperature,
and when the current is relatively high, so that the current
due to the diffusion of the carrier dominates, m 1, and so

the second term in Eq. 1 given in parentheses can be
neglected. Equation 1 can then be simplified to
I Is eqV=kT
The temperature dependence of the saturation current,

Is, is given by
Is AeEg =kT
where Eg is the bandgap energy at T 0 K, and A is a

constant dependent on the geometry and material of the
device. Strictly speaking, A also depends on the temperature. However, the temperature dependency is very weak
compared to the exponential term in Eq. 3. Thus,
I AeqVEg =kT
For a constant current, I, (qVEg)/kT is constant. Thus, the

voltage across a pn junction, V, is a linear function of the
absolute temperature, T. On extrapolating to T 0, then
qV Eg.
The temperature coefficient of V can be derived from
Eq. 4 as

V Eg =q
dV
5
dT I const
T
Since the value of qVEg is always negative, V decreases
with increasing T. For silicon, Eg 1.17 eV, and for T 300
K, V 600 mV, and dV/dT 1.9 mV/K. In actual diodes,
the currentvoltage characteristics have been studied in
detail over a wide temperature range. The forward voltage
exhibits a linear dependence for T > 40 K for a constant
current (11). The observed value of dV/dT in a typical small
signal silicon pn junction diode ranges between 1.3 and
2.4 mV/K for I 100 mA (11). In germanium and gallium
arsenide pn junction diodes, and for silicon Schottky
diodes, the forward voltage exhibits a similar sensitivity
(35).
In most pn junctions, the current through the junction
contains components other than those due to carrier diffusion, and therefore, Eq. 4 does not hold. The base-emitter
pn junction in transistors is advantageous in this respect.
Here, the diffusion component forms a larger fraction of the
total current than that in diodes, even for a diode connection in which the base is connected to the collector. The
nonlinear temperature dependence in the forward voltage
in diode-connected transistors is lower than that of most
diodes (7). Further improvement in linearity is attained
under constant collector current operation, since only the
diffusion component flows to the collector, while other
components flow to the base (12).
From Eq. 2, one can obtain the following expression
ln I ln Is qV=kT
The value of T can be obtained from the gradient of a plot of

ln I versus V, as q and k are known universal constants.
This implies that the currentvoltage characteristics can
be used as an absolute thermometer (6). If ln I is a linear
function of V, only two measurements are required to
determine the gradient. If V1 and V2 are voltages corre-
159
sponding to different current levels, I1 and I2, the difference between these two voltages is calculated using
V1 V2 kT=qlnI1 =I2
Thus, the difference in voltage corresponding to the different current levels for a constant ratio is proportional to the
absolute temperature, without any offset. Using this relationship, a thermometer providing an output proportional
to the absolute temperature can be realized, either by
applying a square wave current to a pn junction (12),
or by using two matched devices operating at different
current levels (13).
Figure 2. A circuit for constant collector current operation in a
sensor transistor.
FUNDAMENTAL CIRCUITS AND DEVICES

A schematic drawing of the fundamental circuit of the
thermometer with a short-circuited transistor or a diode
is shown in Fig. 1. A constant current is applied to the
transistor or diode in the forward bias direction, and the
voltage across the junction is amplified using a differential
amplifier. By adjusting the reference voltage applied to
another input of the differential amplifier, an output voltage proportional to either the absolute temperature in
kelvin or in degrees Celsius or any other desired scale can
be obtained. The operating current of small signal diodes
and transistors is typically 40100 A. If the current
becomes too high, a self-heating error may be produced
due to the power dissipated in the junction. If the current
becomes too small, problems due to leakage and the input
current of the first stage amplifier may become significant
(7).
The nonlinearity in the temperature dependency of the
forward voltage is not a serious problem for most applications, and it can be reduced by appropriate circuit design.
In a Schottky diode, this nonlinearity is < 0.1 K over the
Current
source
Sensor
transistor
or diode
Output
Reference
voltage
input
Figure 1. A fundamental interfacing circuit of a thermometer

making use of a transistor or a diode as a temperature sensor to
provide a voltage output proportional to temperature, with a zero
voltage output at a specific temperature dependent on the
reference voltage selected.
temperature range 65 to 50 8C (5), and a comparable

performance is expected for diode-connected silicon transistors (7). Further improvement in the linearity can be
attained by linearization of the circuit. Linearization using
a logarithmic ratio module reduces the error to <0.05 8C in
the temperature range 65 to 100 8C (7). Linearity is also
improved using a constant collector current, as pointed out
previously. An example of an actual circuit is shown in
Fig. 2. In this circuit, the operational amplifier drives the
base-emitter voltage to maintain a constant collector current. By applying a square-wave current and measuring
the amplitude of the resulting square-wave base-emitter
voltage, a linear output proportional to the absolute temperature is obtained, as expected from Eq. 7(12). Further
improvement in accuracy can be attained by employing a
curve fitting with three-point calibration, the error due to
the nonlinearity can be reduced to 0.01 8C in the temperature range of 50 to 125 8C (14).
Three-terminal monolithic IC temperature sensors that
provide a voltage output proportional to temperature using
the Celsius scale are commercially available, examples
being LM45 (National Semiconductor) and AD22100/
22103 (Analog Devices). The LM45 device operates using
a single power supply voltage in the range 410 V, and
provides a voltage output that corresponds to the temperature in degrees Celsius multiplied by a factor of 10 mV, for
example, 250 mV 25 8C. The AD22100 and AD22103
devices provide a ratiometric output, that is, the output
voltage is proportional to the temperature multiplied by
the power supply voltage. For example, AD22100 has a
sensitivity of 22.5 mV/ 8C giving output voltages of 0.25 V
at 50 8C and 4.75 V at 150 8C when the power supply
voltage is 5.0 V.
Two matched transistors operated using different collector currents can be used to obtain an output proportional
to the absolute temperature (15). The difference in the
base-emitter voltages of the two transistors is a linear
function of temperature, as shown in Eq. 7. Convenient
two-terminal current-output devices using this technique
are commercially available. Figure 3 shows an idealized
scheme representing such devices. If the transistors Q1 and
Q2 are assumed to be identical and have a high commonemitter current gain, their collector currents will be equal,
and will constrain the collector currents Q3 and Q4. If Q3
has r-fold base-emitter junctions, and each one is identical
160
Monolithic temperature sensors that provide a digital

output are also commercially available. For example,
TMP06 (Analog Devices) sensors provide a pulse-width
modulated output. The output voltage assumes either a
high or low level, so that the high period (T1) remains
constant at 40 ms for all temperatures, while the low period
(T2) varies with temperature. In the normal operation
mode, the temperature on the Celsius scale, T, is given by
T 406 731 T1=T2
According to Analog Devices TMP06 data sheet, for an

operating supply voltage between 2.7 and 5.5 V, the absolute temperature accuracy is
1 8C in the temperature
range 070 8C, with a temperature resolution of 0.02 8C.
The National Semiconductor LM75 device is also a
monolithic temperature sensor that provides a digital output. It includes a nine-bit analog-to-digital converter, and
provides a serial output in binary format so that the least
significant bit corresponds to a temperature difference of
0.5 8C.
Newer devices will come along in the future that may be
more appropriate than the ones mentioned here. Information about such devices, together with their data sheets,
will be available from the internet sites of manufactures.
APPLICATIONS
Figure 3. An idealized scheme of a two-terminal IC temperature

sensor that provides a current output proportional to the absolute
temperature.
to that of Q4, the emitter current of a junction in Q3 is 1/r

that of Q4. From Eq. 7, the voltage across resistance R is
obtained from
RI kT=qln r
Thus, the total current, 2I, is proportional to the absolute

temperature. Although the actual components are not
ideal, practical devices are available as monolithic ICs,
such as AD590 and AD592 (Analog Devices) (16). In these
devices, r 8 and R is trimmed to have a sensitivity of
about 1 A/K. The output current is unchanged in the supply-voltage range 4.0 to 30 V. A voltage output proportional
to the absolute temperature can be obtained by connecting
a resistor in series with the ICs. For example, a sensitivity
of 1 mV/K is obtained by connecting 1 kV resistor in series.
By trimming the series resistor, the error in temperature
reading can be adjusted to zero at any desired temperature.
After trimming, the maximum error depends on the range
in temperature under consideration. For example, a maximum error of <0.1, 0.2, and 0.3 8C is obtained for temperature ranges of 10, 25, and 50 8C, respectively (17).
Although thermistors are still widely used for thermometry in the medical field, IC temperature sensors have
potential advantages over thermistors. Integrated circuit
sensors can be fabricated using IC technology encompassing interfacing electronics on a single IC chip, and many
general purpose IC temperature sensors are now commercially available.
Current-output-type IC temperature sensors, such as
AD590, are convenient for use as thermometer probes for
body core and skin temperature measurements. Figure 4
shows a scheme for such a simple thermometer. According
to the manufacturers data sheet, although the sensitivity
and zero offset are adjustable independently in this circuit,
an accuracy of 0.1 8C is attainable with L- or M-grade
AD590 devices using a single-trim calibration if the temperature span is 10 8C or less. If a regulating resistor is
included in the probe, interchangeability can be realized.
Because of the current output capacity, the resistance of
Figure 4. A simple thermometer that makes use of a twoterminal current output-type IC temperature sensor.
Figure 5. A multiplexing scheme for a current output-type IC

temperature sensor.
the cable or connector does not affect the temperature

measurement.
This type of device is also convenient for temperature
measurements at many other points, especially when the
output data are processed using a PC. All the sensors can
be connected to a single resistor, as shown in Fig. 5, and by
switching the excitation the outputs from each sensor can
be multiplexed. To calibrate each sensor individually, all
the sensors are maintained at an appropriate temperature,
together with a standard thermometer. The outputs from
each sensor as well as that from a standard thermometer
are input into a PC. Then, the temperature offsets for each
sensor can be stored, and all the measurement data can be
corrected using these correction factors. Two-point calibration is also realized by using data at two known temperatures. A matrix arrangement of the sensors can be formed
using two decoder drivers.
161
Temperature measurements at many different points

can be performed easier using IC temperature sensors that
generate serial digital outputs, such as TMP05/TMP06.
Connecting these devices as shown in Fig. 6 allows for the
realization of a daisy chain operation. When a start pulse is
applied to the input of the first sensor, the temperature
data from all the sensors is generated serially, so that the
temperatures of each sensor are represented in a ratiometric form, which is the ratio of the duration of the high
and low output levels for each period. It is a remarkable
advantage of this sensor that a thermometer can be realized without using any analogue parts.
An important application of IC temperature sensors is
the monitoring of CPU temperatures to protect a system
from overheating. The temperature of a CPU chip can be
detected by a pn junction fabricated on the same silicon
chip as the CPU, as shown in Fig. 7. The advantage of
fabricating the temperature sensor on the CPU chip is to
make the temperature measurement accurate enough and
to minimize the time delay due to heat conduction so as to
prevent overheating. The CPU can be protected from overheating by controlling a cooling fan or by slowing down the
clock speed. Interfacing devices for this purpose are
commercially available. For example, the MAX6656 (Dallas Semiconductor) device can detect temperatures at three
locations, such as the CPU, the battery, and the circuit
board, and the output can be used to control a cooling fan.
To control the clock frequency, a specially designed frequency generator can be used. For example, the AV9155
(Integrated Circuit Systems) device allows for a gradual
transition between frequencies, so that it obeys the CPUs
cycle-to-cycle timing specifications.
FUTURE
It is 25 years since convenient IC temperature sensors
were introduced for scientific and industrial temperature
measurements. In medicine, the application of this type of
sensor is in its infancy. There are many applications where
VDD
TMP05
#1
Start pulse
TMP05
#2
TMP05
#N
Output
GND
(a)
#1
T1
#N
#2
T2
Celsius temperature = 406 (731 (T1/T2 ))

(b)
Figure 6. (a) The connecting scheme for a

daisy chain operation of a serial-digitaloutput-type temperature sensor, and (b)
the output waveform. The temperature
using the Celsius scale at each sensor
can be determined from the ratio of the
duration of the highest and lowest points
in each cycle.
162
Figure 7. A scheme for monitoring the temperature of a CPU to

protect it from overheating by fan control or by slowing down the
clock.
these sensors can be used effectively, and undoubtedly

their use will be wide spread in the near future.
Digital output IC temperature sensors show the most
promise. Using such sensors, thermometers can be made
without using analog components, and digital signals are
convenient when a photocoupler is used for isolation.
Medical thermometry requires a relatively high degree
of accuracy within a narrow temperature range. An absolute accuracy of 0.1 8C is required for body temperature
measurements. However, this is hard to attain without
individual calibration using most temperature sensors.
While adjustment of the trimmer resistor has been used
in many thermometer units, correcting data using a PC
employing initially obtained correction factors will be much
simpler, especially when many sensors are used, and digital output IC temperature sensors are advantageous for
such a purpose.
Fabricating different types of sensors, such as force and
temperature sensors, in one chip, and then applying them
in robot hands to mimic all the sensing modalities of human
skin, is another promising field. In such applications, the
digital output capability will be a great advantage.
BIBLIOGRAPHY
1. Mackay RS. Endoradiosonde. Nature (London) 1957;179:
12391240.
2. Harris H. Concerning a thermometer with solid-state diodes.
Sci Amer 1961;204(6):192.
3. MacNamara AG. Semiconductor diodes and transistors
as electrical thermometers. Rev Sci Instrum 1963;33: 1091
1093.
4. Cohen BG, Snow WB, Tretola AR. GaAs p-n junction diodes for
wide range thermometry. Rev Sci Instrum 1963; 34:1091
1093.
5. Griffiths B, Stow CD, Syms PH. An accurate diode thermometer for use in thermal gradient chambers. J Phys E 1974;
7:710714.
6. Felimban AA, Sandiford DJ. Transistors as absolute thermometers. J Phys E 1974;7:341342.
7. Davis CE, Coates PB. Linearization of silicon junction characteristics for temperature measurement. J Phys E
1977;10:613619.
8. ONeil P, Derrington C. Transistorsa hot tip for accurate
temperature sensing. Electronics 1979;52(21):137141.
9. Sah C, Noyce RN, Shockley W. Carrier generation and recombination in p-n junctions and p-n junction characteristics. Proc
IRE 1957;45:12281243.
10. Sah C. Effect of surface recombination and channel on p-n
junction transistor characteristics. IRE Trans Electron
Devices 1962;ED9:94108.
11. Sclar N, Pollock DB. On diode thermometers. Solid State
Electron 1972;15:473480.
12. Verster TC. p-n junction as an ultralinear calculable thermometer. Electron Lett 1968;4:175176.
13. Ruhle RA. Solid-state temperature sensor outperforms previous transducers. Electronics 1975;48(6):127180.
14. Ohte A, Yamagata M. A precision silicon transistor thermometer. IEEE Trans Instrum Meas 1977;IM-26:335341.
15. Vester TC. Dual transistor as thermometer probe. Rev Sci
Instrum 1969;40:174175.
16. Timko MP. A two-terminal IC temperature transducer. IEEE
J Solid-State Circuits 1976;SC-11:784788.
17. Sheingold DH, editor. Transistor Interfacing Handbook, A
Guide to Analog Signal Conditioning. Norwood, MA: Analog
Devices; 1980. p 153177.
Further Reading
Sze SM. Semiconductor DevicesPhysics and Technology. New
York: John Wiley & Sons; 1985.
berg PA. Biomedical Transducers and
Togawa T, Tamura T, O
Instruments. Boca Raton, FL: CRC Press; 1997.
Moore BD. IC temperature sensors find the hot spot. EDN July 2/
98, 1998; 99110.
Frank R. Semiconductor junction thermometers. In: Webster JG,
editor. The Measurement, Instrumentation, and Sensors
Handbook. Boca Raton, FL: CRC Press; 1999. p 32/7432/87.
See also CAPACITIVE
MICROSENSORS FOR BIOMEDICAL APPLICATIONS; ION-
SENSITIVE FIELD EFFECT TRANSISTORS; THERMOMETRY.
INTERFERONS. See IMMUNOTHERAPY.

INTERSTITIAL HYPERTHERMIA. See
HYPERTHERMIA,
INTERSTITIAL

PETER WELLER
City University
London, United Kingdom
DARREN MORROW
Royal Adelaide Hospital
Adelaide, Australia
INTRODUCTION
The heart is a pump made of cardiac muscle or myocardium. It has four pumping chambers, namely, a right and
left atrium and a right and left ventricle. The atria act as
primer pumps for the ventricles. The right ventricle pumps
deoxygenated blood returning from the body through the
pulmonary artery and into the lungs. This is called the
pulmonary circulation. The left ventricle pumps oxygenated blood returning from the lungs through the aorta
and into the rest of the body. This is called the systemic
circulation.
The heart also has four one-way valves that prevent the
backward flow of blood. The tricuspid valve lies between
the right atrium and right ventricle while the pulmonary
valve lies between the right ventricle and the pulmonary
artery. Similarly, the mitral valve lies between the left
atrium and the left ventricle while the aortic valve lies
between the left ventricle and the aorta.
163
ventricle. The period between the closure of the aortic valve

and opening of the mitral valve is called isovolumetric
relaxation.
The left atrium contracts and relaxes just before the left
ventricle. This boosts the blood flow from the left atrium
into the left ventricle.
These events are mirrored in the right ventricle and
right atrium. However, the pressures in the pulmonary
circulation are much lower than the pressures in the
systemic circulation.
The movements of the chambers, valves and blood can
be imaged noninvasively using ultrasound and this is
called an echocardiogram.
Electrical Events
Figure 1. The relationship between the aortic pressure (A

dashed line), the ventricular pressure (B solid line), the
electrocardiogram (C) and the heart sounds (D). Region
1 isovolumetric contraction and Region 2 isovolumetric
relaxation, Region 3 (green) tension time index (TTI) and
Region 4 (yellow) diastolic pressure time index (DPTI). S1
represents the closing of mitral and tricuspid valves, S2
represents the closure of aortic and pulmonary valves.
CARDIAC CYCLE
The heart pumps rhythmically. The cardiac cycle is the
sequence of events that take place in the heart during one
heartbeat (Fig. 1). Thus, the duration of the cardiac cycle
varies inversely with the heart rate. At a typical resting
heart rate of 60 beats per minute (bpm), the cardiac cycle
lasts 1 s or 1000 ms.
The rhythmical pumping of the heart is caused by waves of

electrical impulses that spread through the myocardium
from the atria to the ventricles. A recording of these waves
is called an electrocardiogram (ECG). The P wave represents atrial contraction. The R wave represents ventricular
contraction and signals the beginning of systole. The T
wave represents ventricular relaxation and signals the
beginning of diastole.
Acoustic Events
The opening and closing of the valves in the heart creates
sounds that can be heard at the surface of the chest using a
stethoscope. A recording of these sounds is called a phonocardiogram. The first heart sound (S1) represents closure of the mitral and tricuspid valves and signals the
beginning of systole. The second heart sound (S2) represents closure of the aortic and pulmonary valves and
signals the beginning of diastole.
MYOCARDIAL OXYGEN BALANCE
Mechanical Events
One cardiac cycle consists of a period of contraction called
systole followed by a period of relaxation called diastole.
The duration of systole, called the systolic time interval
(STI), is relatively constant, but the duration of diastole,
called the diastolic time interval (DTI), varies with the
heart rate. Thus, when the heart rate increases, the DTI
shortens.
When the left ventricle contracts, the pressure in the left
ventricle rises above the pressure in the left atrium and the
mitral valve closes. Soon afterward, the pressure in the left
ventricle rises above the pressure in the aorta and the
aortic valve opens. Blood flows from the left ventricle into
the aorta. The period between closing of the mitral valve
and opening of the aortic valve is called isovolumetric
contraction.
When the left ventricle relaxes, the pressure in the left
ventricle falls below the pressure in the aorta and the aortic
valve closes. This causes a momentary drop in pressure in
the aorta called the dichrotic notch. The period between the
opening and closing of the aortic valve is called ventricular
ejection. Soon afterward, the pressure in the left ventricle
falls below the pressure in the left atrium and the mitral
valve opens. Blood flows from the left atrium into the left
The systemic circulation delivers oxygenated blood to the

body. Body tissues use oxygen to generate energy from the
oxidation of fuels. All tissues, including the myocardium,
need energy to function. The net delivery of oxygen to the
myocardium is called the myocardial oxygen balance.
MOB MOS MOD
where MOB myocardial oxygen balance, MOS myocardial oxygen supply, MOD myocardial oxygen demand. In
the healthy heart the myocardial oxygen balance is positive, that is supply exceeds demand. In the failing heart the
balance can be negative, that is demand exceeds supply.
Myocardial Oxygen Supply
The main blood supply of the myocardium comes from the
two coronary arteries and their branches. The small
amount of blood that reaches the myocardium transmurally from within the chambers of the heart is insignificant.
The coronary arteries arise from the aorta just beyond
the aortic valve and ramify within the myocardium.
Myocardial oxygen supply depends on the coronary blood
flow and the amount of oxygen that can be extracted from
the blood.
164
When the heart contracts the coronary arteries are

compressed and the coronary blood flow is decreased.
The net driving force for coronary blood flow is called
the coronary perfusion pressure.
CPP AP VP
where CPP coronary perfusion pressure, AP aortic
pressure, VP ventricular pressure. The coronary circulation is unique because more blood flows during diastole
when the ventricular pressure is low than during systole
when the ventricular pressure is high. Thus, the coronary
blood flow depends on the coronary perfusion pressure, the
diastolic time interval and the patency of the coronary
arteries. Myocardial oxygen supply is represented by the
area between the aortic pressure wave and the left ventricular pressure wave, called the diastolic pressure time
index (DPTI).
Myocardial Oxygen Demand
The myocardium uses energy to perform the work of pumping. The work performed by the heart can be estimated by
the mean aortic blood pressure multiplied by the cardiac
output. Myocardial oxygen demand depends on the heart
rate, the systolic wall tension and the cardiac contractility.
Systolic wall tension is developed during isovolumetric
contraction and depends upon the preload, the afterload
and the wall thickness. The preload is the degree to which
the left ventricle is filled before it contracts, that is the left
ventricular end diastolic volume. The afterload is the
pressure in the aorta or the systemic vascular resistance
against which the left ventricle contracts. Myocardial oxygen demand is represented by the area under the left
ventricular pressure curve, called the tension time index
(TTI).
The myocardial oxygen balance is represented by the
ratio DPTI:TTI.
THE PATHOPHYSIOLOGY OF LEFT VENTRICULAR
PUMP FAILURE
When the left ventricle begins to fail as a pump, the cardiac
output falls. Compensatory physiological mechanisms
bring about an increase in left ventricular end diastolic
volume, heart rate, and systemic vascular resistance. The
result is an increase in preload and afterload with a
decrease in coronary blood flow. Thus, the myocardial
oxygen demand increases while the myocardial oxygen
supply decreases. There may come a point when demand
exceeds supply resulting in a negative myocardial oxygen
balance. The left ventricle is then deprived of oxygen and
cannot generate sufficient energy to do the work required
of it. The pump failure is therefore exacerbated and this
can precipitate a downward spiral of decline eventually
ending in death. The therapeutic goal is to reverse this
decline and help the failing left ventricle to recover by
restoring a positive myocardial oxygen balance. Diuretics
to decrease the preload, inotropic drugs to increase the
myocardial contractility and vasodilators to decrease the
preload and afterload are the mainstay of treatment. However, in the most severely ill patients, pharmacological
Figure 2. The relationship between the ventricular pressure with

counterpulsation (A solid line), without counterpulsation (Bdashed line), IABP balloon inflation (C) and ventricular systole
(D). Region 1 (green) systolic unloading and Region 2
(yellow) diastolic augmentation.
measures alone may be insufficient and it is in these

extreme circumstances that counterpulsation therapy
may be effective.
THE PRINCIPLE OF COUNTERPULSATION
The principle of counterpulsation is the incorporation of an
additional pump into the systemic circulation in series with
the left ventricle. The pump is operated in synchrony, but
out of phase, with the cardiac cycle. Pump systole occurs
during ventricular diastole and pump diastole occurs during ventricular systole.
The primary physiological effects of counterpulsation
are twofold (Fig. 2): A decrease in the aortic pressure
during systole (called systolic unloading). This is evidenced
by a decrease in the end diastolic pressure (EDP), the peak
systolic pressure (PSP) and the mean systolic pressure
(MSP). An increase in the aortic pressure during diastole
(called diastolic augmentation). This is evidenced by an
increase in the mean diastolic pressure (MDP).
Systolic unloading reduces the work of the left ventricle
because it pumps against a lower pressure. This decreases
myocardial oxygen demand. Diastolic augmentation
increases coronary blood flow because it increases the
coronary perfusion pressure. This increases myocardial
oxygen supply. Thus, the myocardial oxygen balance is
improved.
Among the secondary physiological effects of counterpulsation are increases in the stroke volume (SV, the
volume of blood pumped with each heartbeat), the CO
(equal to the SV multiplied by the heart rate) and the
blood flow to the other vital organs.
HISTORICAL PERSPECTIVE
Counterpulsation was first described in theory in 1958 by
Harken (1). It was to be achieved by cannulating the
femoral arteries, rapidly withdrawing a set volume of blood
during systole and rapidly reinfusing the same volume of
blood during diastole. Clauss et al. (2) reported this in
clinical practice in 1961 but it was unsuccessful because
the rapid movements of blood were difficult to implement
and caused severe hemolytic damage to the red blood cells.
In 1958, Kantrowitz and McKennin (3) described counterpulsation achieved by wrapping a part of the diaphragm
around the thoracic aorta and stimulating the phrenic
nerve, causing contraction of the diaphragm, during diastole. Moulopoulos et al. (4) and Clauss et al.(5) described
counterpulsation achieved by intra-aortic balloon pumping
in 1962. Operative insertion of the balloon through a surgically exposed femoral artery was necessary. It was inflated
during diastole and deflated during systole. In 1968, Kantrowitz et al. (6) reported a successful clinical study.
In 1963, Dennis et al.(7) described external counterpulsation achieved using a pneumatic compression garment that enclosed the legs and lower torso. It was inflated
during diastole and deflated during systole. It was reported
to be as successful as the IABP in clinical studies but it is
not commonly used. Kantrowitz et al. (8) described counterpulsation achieved by a permanently implantable intraaortic balloon in 1972. It was unsuccessful in clinical
practice because there remained the need for a connection
to an external pump and this provided a portal of entry for
infection.
In 1979, following the development of thinner catheters,
percutaneous insertion of the intra-aortic balloon through
a femoral artery puncture was introduced. This could be
performed at the bedside and avoided the need for a
surgical operation in most patients. Consequently, intraaortic balloon pumping became the most widely adopted
method of counterpulsation.
CLINICAL APPLICATIONS
Indications
The IABP was first used clinically in 1968 by Kantrowitz to
support patients with cardiogenic shock after acute myocardial infarction(9). During the 1970s the indications
broadened (Table 1) and by 1990 70,000 pump procedures
were performed worldwide each year (10) although there is
wide variation between different countries and centres.
The IABP support has been used successfully in patients
with left ventricular failure or cardiogenic shock from
many causes including myodarditis, cardiomyopathy,
severe cardiac contusions and drug toxicity but the commonest are myocardial infarction and following cardiac
surgery. The trend has been a move away from hemodynamic support in pump failure towards the treatment, and
even prophylaxis of, acute myocardial ischaemia. Patients
can be maintained on the IABP for hours, days or even
weeks, particularly when used as a bridge to cardiac
transplantation or other definitive treatment (11). Of
those who survive to hospital discharge, long-term survival
is satisfactory (12).
An early series of 747 IABP procedures in 728 patients
between 1968 and 1976 was reported by McEnany et al.
165
(13). Over the course of the study, they observed that

cardiogenic shock or chronic ischaemic left ventricular
failure as the indication for IABP fell from 79 to 26% of
patients whilst overall in-hospital survival rose from 24 to
65% of patients. They also noted an increase from 38 to 58%
of patients undergoing cardiac surgery following IABP
insertion. They postulated that broadened indications
for, and earlier insertion of, the IABP together with more
aggressive surgical treatment of any underlying cardiac
lesion led to the improvement in survival. In the later
Benchmark Registry of nearly 17,000 IABP procedures
performed in 203 hospitals worldwide between 1996 and
2000, the main indications were support for coronary
angioplasty (21%), cardiogenic shock (19%), weaning from
cardiopulmonary bypass (16%), preoperative support in
high risk patients (13%), and refractory unstable angina
(12%) (14). The overall in-hospital mortality was 21%.
High risk patients undergoing cardiac surgery may
have a better outcome if treated preoperatively with IABP
therapy. In a series of 163 patients with a left ventricular
ejection fraction of <0.25 and undergoing coronary artery
bypass grafting (CABG), the 30 day mortality was reduced
from 12 to 3% (15). Similar results were obtained in a small
randomized study (16). In a series of 133 patients who
underwent CABG off cardiopulmonary bypass between
2000 and 2003, the use of adjuvant preoperative IABP
therapy in the 32 highest risk patients led to outcomes
comparable with the lower risk patients (17). The use of
IABP therapy to improve outcome after coronary angioplasty for acute myocardial infarction remains controversial. Early studies suggested an improved outcome (1820),
but two recent large randomized trials have shown no
benefit in haemodynamically stable patients (21,22). A
report from the SHOCK Trial Registry showed that the
in-hospital mortality in patients with cardiogenic shock
after acute myocardial infarction could be reduced from
77% to 47% by combined treatment with thrombolysis and
IABP, particularly when followed by coronary revascularization (23).
IABP therapy is used infrequently in children, who
commonly suffer from predominantly right ventricular
failure associated with congenital heart disease. The
greater elasticity of the aorta may limit diastolic augmentation and the more rapid heart rate may make ECG
triggering difficult. Echocardiographic triggering has
been used as an effective alternative (24,25). Survival
rates of 57% (26) and 62% (27) have been reported in small
series of carefully selected patients.
Contraindications
The only absolute contraindications to IABP therapy are
severe aortic regurgitation and aortic aneurysm or dissection. In patients with severe aorto-iliac vascular disease
Table 1. Indications for IABP Therapy

Left ventricular failure or cardiogenic shock
Refractory unstable angina or ischaemic ventricular arrhythmias
Weaning from cardiopulmonary bypass
Bridge to cardiac transplantation
Preoperative support before cardiac or non-cardiac surgery

Adjunct to coronary angioplasty or thrombolyis
Adjunct to off-bypass cardiac surgery
166
Table 2. Complications of IABP Therapy

Vascular
Balloon-Related
Other
Hemorrhage
Aortoiliac dissection
or perforation
Gas embolism
Entrapment
Infection
Thrombocytopenia
Paraplegia
Limb ischaemia
Visceral ischaemia
the balloon should not be inserted through the femoral

artery.
Complications
The IABP therapy continues to cause a significant number
of complications (Table 2), but serious complications are
uncommon and directly attributable deaths are rare.
Nevertheless, some have argued against its indiscriminate
use, feeling that for many patients the risks outweigh the
benefits. Kantrowitz reported rates of 41 and 4% for minor
and major complications, respectively, in his series of 733
patients. Of these, 29% were vascular (including 7%
hemorrhagic) and 22% were infections (28). Vascular complications include haemorrhage from the insertion site and
lower limb ischaemia caused by the balloon catheter or
sheath occluding the iliac or femoral artery. The vascular
status of the lower limbs should be observed closely in
patients on IABP therapy. Ischaemia may resolve when the
catheter or sheath is removed but surgical intervention
including femoral thromboembolectomy, femorofemoral
bypass or even amputation is required in up to half of
cases (29).
Several risk factors for vascular complications have
been identified. They are female gender, diabetes, hypertension, peripheral vascular disease, obesity, old age,
sheathed insertion, percutaneous insertion, and insertion
via the femoral artery compared to directly into the ascending aorta (13,14,19,2830). In one study of patients with
peripheral vascular disease, the rate of vascular complications was 39% for percutaneous insertion compared to 18%
for open insertion (29).
Complications caused by perforation of the balloon are
rare, but potentially serious. Embolization of the helium
shuttle gas can result in stroke or death. Coagulation of
blood within the balloon can result in balloon entrapment.
In this situation the instillation of thrombolytic agents may
allow the balloon to be retrieved percutaneously but otherwise open surgery is required. It is therefore mandatory to
remove the balloon immediately if any blood is detected
within the pneumatic system.
Thrombocytopenia (a reduction in the number of platelets) developed in one-half of patients but they rapidly
recovered when the balloon was removed (31).
EQUIPMENT FOR CLINICAL APPLICATION
The IABP consists of a balloon catheter and movable drive
console. A monitor on the drive console displays the arterial
pressure wave and the ECG. Commercial consoles have
Figure 3. A commercial IABP device. (Courtesy of Datascope

Corp.)
controls that allow the operator to select the assist ratio

and trigger mode and adjust the timing of inflation and
deflation and the inflation volume of the balloon. The drive
console also contains a helium tank for balloon inflation
and a battery as a backup power source in the event that
the mains electricity supply is interrupted. In common
with medical equipment the IABP console conforms to
international safety standards. Figure 3 shows a current
commercial model.
The balloon is made of inelastic polyurethane and is
cylindrical in shape. Balloons are available in volumes
from 25 to 40 cm3 and the correct size is selected according
to the height of the patient. The balloon is mounted at the
end of a double-lumen catheter. Modern catheters have an
outer diameter of 78 French gauge. The inner lumen is
open at the tip to allow insertion over a guidewire and
direct measurement of the aortic blood pressure after the
guidewire is removed. The outer lumen forms a closed
system connecting the balloon to a pneumatic pump chamber within the drive console. Two views of a current balloon
catheter are shown in Fig. 4.
The balloon catheter is most commonly inserted percutaneously through the femoral artery in the groin over a
guidewire using a traditional or modified Seldinger technique. An intra-arterial sheath is used to secure and
167
Monitor
Console
Figure 4. An IABP catheter. (Courtesy of Datascope Corp.) In
both inflated (top image) and uninflated (bottom image) modes.
protect the access site before insertion of the balloon catheter. However, because the sheath has a larger outer diameter than the balloon catheter it can increase the risk of
vascular complications and sheathless insertion has now
been introduced. Under fluoroscopic guidance the balloon
is positioned in the descending thoracic aorta just beyond
the origin of the left subclavian artery. Alternatively, the
balloon can be inserted through the iliac, axillary or subclavian arteries or directly into the ascending aorta during
open surgery.
A shuttle gas is pumped back and forth between the
pump chamber and the balloon to cause inflation and
deflation. The ideal shuttle gas would have a low density
combined with a high solubility in blood. A dense gas is
slow to move along the catheter. This introduces a significant delay between the opening of the valves in the pump
chamber and the inflation or deflation of the balloon making correct timing more difficult to achieve. An insoluble
gas is unsafe if the balloon was to leak or burst allowing it
to escape into the blood. There it may form bubbles leading
to potentially fatal gas embolism. Originally, carbon dioxide was used as the shuttle gas. It is a dense gas but
dissolves easily in blood. More recently, helium has been
used as the shuttle gas. It is a less dense gas but dissolves
less easily in blood.
Pumping is initiated with an assist ratio of 1:2, which
means that one in every two heartbeats is assisted. This
allows the arterial pressure wave of each assisted beat to be
compared with an unassisted beat to facilitate correct
timing.
Pumping is continued with an assist ratio of 1:1, which
means that every beat is assisted. As the patient recovers,
they can be weaned from the IABP by periodically decreasing the assist ratio until pumping is eventually discontinued.
Weaning can also be achieved by periodically decreasing
the inflation volume of the balloon. However, this can lead
to problems with blood clotting in the folds of the underinflated balloon and it is not commonly used.
CONTROL OF IABP
The inflation/deflation cycle of the intra-aortic balloon
pump is controlled by a closed loop circuit as illustrated
in Fig. 5.
The physiological variables required for determining
triggering are measured, filtered and converted into digital
signals. These are used in enable the control strategy to
determine the appropriate inflation and deflation times of
the balloon. This information is then passed to the pneumatic circuit for balloon operation. The results of this
strategy are then feed back to the console via a new set
of patient variables. The actions of the controller are dis-
Control strategy
from user
Inflation &
Deflation
strategy
Information for
displaying to user
Control Strategy
Enhanced physiological
measurements
Pneumatic
System
Analogue to Digital
Converters and Filters
Measurements of
physiological variables
Balloon
inflation &
deflation
Figure 5. Closed-loop circuit for IABP control. The arrows

indicate the flow of information for controlling the IABP.
played on the console monitor. Additionally the control

strategy can be set by the clinician.
TRIGGERING
The drive console requires a trigger signal to synchronise
with the cardiac cycle. The trigger signal indicates the start
of ventricular systole.
Commercial consoles have several modes of triggering:
The ECG trigger, where the trigger signal is the R wave of
the ECG. This is the most commonly used mode of triggering. Blood pressure trigger, where the trigger signal is the
upstroke of the arterial pressure wave. This is used when
the ECG signal is too noisy to allow reliable R wave
recognition. Pacer trigger, where the trigger signal is
the pacing spike of the ECG. This is used when the patient
has an implanted cardiac pacemaker. Internal trigger,
where the trigger signal is generated internally by the
console at a set rate. This is used during cardiac surgery
when the heartbeat is temporarily arrested.
TIMING
The safety and efficacy of balloon pumping is dependent
upon the correct timing of balloon inflation and deflation
during the cardiac cycle. Inflation should occur during the
isovolumetric relaxation period of ventricular diastole. Too
early inflation is unsafe because it overlaps the ejection
period of ventricular systole, impedes ejection and
increases the work of the heart. Too late inflation is less
effective because it reduces diastolic augmentation.
168
Safe and effective inflation timing is fairly easy to

achieve because the systolic time interval is relatively
constant, regardless of heart rate or rhythm. The operator
is trained to adjust the time of inflation until it visibly
corresponds with the dichrotic notch on the arterial pressure wave display.
Deflation should occur at the end of ventricular diastole
or during the isovolumetric contraction period of ventricular
systole. Too late deflation is unsafe because it overlaps the
ejection period of ventricular systole, impedes ejection and
increases the work of the heart. Too early deflation is less
effective because it reduces diastolic augmentation.
Safe and effective deflation timing is more difficult to
achieve because the length of diastole is variable, depending on the heart rate and rhythm. The operator is trained to
adjust the time of deflation until it visibly reduces the EDP
and PSP on the arterial pressure wave display. Commercial consoles have two modes of timing:
Conventional Timing
Under conventional timing, the balloon is inflated and
deflated at set time intervals after the R wave is detected
(called manual timing). This copes badly with alterations
in heart rate of >10 bpm. The problem is that when the
heart rate increases, the diastolic time interval shortens
and the balloon now deflates too late. Conversely, when the
heart rate decreases, the diastolic time interval lengthens
and the balloon deflates too early.
Conventional timing can be improved by predicting the
duration of the current cardiac cycle by averaging the RR
interval of the previous 520 heartbeats. The balloon is
then deflated at a set time interval before the next R wave
is predicted to arrive (called predictive timing). This copes
fairly well with alterations in heart rate and is the most
commonly used mode of timing.
Both manual and predictive timing cope badly with
alterations in heart rhythm, when the length of diastole
can vary from beat to beat in an entirely unpredictable
way. Unfortunately, cardiac arrhythmias such as atrial
fibrillation and frequent ectopic beats are common in these
patients making optimal timing difficult to achieve.
Real Timing
Under real timing (also called R wave deflation), the balloon is deflated when the R wave is detected and inflated a
set time interval later. This copes very well with alterations in heart rate and rhythm but it tends to cause too late
deflation. The problem is that when the R wave is detected
there may be insufficient time to fully deflate the balloon
before overlapping the ejection period.
The R wave deflation is usually used as a safety mechanism in conjunction with conventional timing. It ensures
that the balloon is deflated if an R wave arrives unexpectedly due to a sudden change in heart rate or rhythm.
OPTIMIZATION
As was seen above the standard timing strategies both
suffer from problems in certain scenarios and thus control
of IABPs are not efficient in providing the best patient

treatment. A natural solution to this is to develop timing
strategies that optimize the balloon inflation regime
according to the current patient condition. Several teams
have reported work in this area.
Jaron et al. in 1979 (32) developed a multielement
mathematical model of the canine circulation and the
IABP. They expressed inflation and deflation times in
terms of the total duration of inflation (DUR) and the time
from the R wave to the middle of pump systole (TMPS).
Duration of inflation and TMPS were expressed in terms of
percentages of the duration of the cardiac cycle.
The model was validated by comparison with anesthetized dogs. They varied DUR and TMPS and measured the
effects on EDP, MDP, and CO. Each of the three dependent
variables (z axis) were plotted against the two independent
variables (x and y axes) to create a three-dimensional (3D)
urface. In general, there was considerable similarity
between the surfaces obtained from the model and from
the dogs. In particular, the similarity was closer for measurements of pressure (EDP, MDP) than for measurements
of flow (CO).
Their results indicated that the locations of the desired
optima for EDP (75% DUR, 45% TMPS) & CO (55% DUR,
75% TMPS) did not coincide. Furthermore, some combinations of DUR and TMPS within the range used clinically
produced detrimental effects. Because not all variables
could be optimised at same time, they suggested that the
choice of timing settings involved balancing the clinical
needs of the patient.
Jaron et al. in 1983 (33) subsequently developed a
lumped model of the canine circulation and the IABP. It
was validated by comparison with their previous model.
They varied the timing of inflation and deflation, the speed
of inflation and deflation and the volume of the balloon. The
speed was either fast or slow, taking 7% or 33% of the
duration of the cardiac cycle, respectively. The fast speed
represented the ideal console with near instantaneous
balloon inflation and deflation. The slow speed represented
commercial consoles with significant inflation and deflation delay due to the movement of the shuttle gas. They
measured the effects on EDP, SV, and coronary blood flow.
Inflation at end systole maximized SV and coronary
blood flow for fast and slow speeds. Deflation at end diastole minimized EDP for fast speeds. At slow speed, deflation timing involved a trade-off between decreased EDP
with early deflation and increased SV and coronary blood
flow with late deflation. The overall benefit of the IABP was
greater with fast speeds than slow speeds. It was also
proportional to balloon volume.
In later experiments (34), they classified dependent
variables as either internal, reflecting myocardial oxygen
demand, or external, reflecting myocardial oxygen supply.
Internal variables measured were TTI and EDP. External
variable measured were SV and MDP. They showed that
early deflation minimizes internal variables while late
deflation maximizes external variables.
Niederer and Schilt in 1988(35) used a mechanical mock
circulation and a mathematical model to investigate then
influence of timing of inflation and deflation, speed of
inflation and deflation and balloon volume on the efficacy
of the IABP. Timing was again expressed in terms of

percentage of cardiac cycle duration. The default settings
of the model were fast inflation at 30% time, fast deflation
at 90% time and a volume of 40 cm3. These were found to
produce an increase in SV of 25%, a decrease in left
ventricular systolic pressure of 10% and an increase in
aortic diastolic pressure of 50%.
The time of inflation was varied between 20% and 50%.
This had little effect on SV when the speed was fast, but
inflation after 30% caused a slight decrease in SV compared
to default when the speed was slow. A time of deflation
before 80% caused a slight decrease in SV compared to
default. Fast inflation and deflation speeds caused opening
and closing shock waves that could be harmful were they to
occur in humans. Balloon volume was varied between 10
and 50 mL. There was a nonlinear relationship with SV,
but 40 mL was adequate for optimal performance in the
mock circulation.
Barnea et al. in 1990 (36) developed a sophisticated
computer simulation of the normal, failing and IABPassisted failing canine circulation. It included simple physiological reflexes involved in the regulation of the cardiovascular system. The failing heart was simulated by
reducing the contractility of the normal heart. Myocardial
oxygen supply and demand were calculated from the
model. They were balanced in the normal circulation
and imbalanced in the failing circulation. IABP assistance
of the failing circulation was shown to restore the balance.
Sakamoto et al. in 1995 (37) investigated the effect of
deflation timing on the efficiency of the IABP in anaesthetized dogs. They compared a deflation time before the R
wave (during late diastole) with deflation times after the R
wave (during isovolumetric contraction). Deflation during
the middle of isovolumetric contraction was the most effective in obtaining optimal systolic unloading.
Morrow and Weller (38) successfully used genetic algorithms and the fitness function proposed by Kane et al.(39)
to evolve a fuzzy controller that optimized cardiac assistance in a computer simulation of the IABP-assisted failing
heart. The inputs were MDP and PSP and the output was
deflation time.
Automatic Control
Kane et al. in 1971 (39) proposed a performance index or
fitness function that reflected the overall benefit of IABP
assistance at different timing combinations. Inflation and
deflation times were expressed in terms of the delay before
inflation after the R wave and the duration of inflation. The
function included weighted MDP, MSP, and EDP.
Fitness k1 MDP k2 MSP k4 dk3 EDP2
100
100
; k2
; k3 EDP0 ;
MDP0
MDP0

2
1 ifk3 EDP < 0
1
; d
k4 500
k3
0 otherwise
MDP0 unassisted MDP; EDP0 unassisted EDP
where;
k1
It was tested in a mechanical mock circulation and

fitness was found to be a unimodal function of delay and
169
duration. An automatic controller was developed that used

a gradient descent search algorithm to improve the fitness
by adjusting the delay and duration at each heartbeat. The
performance of the controller was compared with the performance of R wave deflation in simulated cases of heart
failure. The controller was considerably better in moderate
heart failure and marginally better in severe failure. In
severe failure the trade-off between systolic unloading or
diastolic augmentation was particularly apparent. In moderate failure, the fitness function emphasized systolic
unloading (early deflation) while in severe failure it
adapted to emphasize diastolic augmentation (late deflation). Thus, in severe failure the controller tended to
simulate R wave deflation.
Martin and Jaron in 1978 (40) developed a manual
controller for the IABP that allowed DUR and TMPS to
be adjusted. It was tested successfully on anesthetized dogs
and was capable of linking to a computer for automatic
control.
Jaron et al. in 1985 (34) suggested that SV and TTI
index were suitable choices for a fitness function for the fine
adjustment of inflation time. Both MDP and EDP were
suitable choices for the adjustment of deflation time. However, later work showed that PSP correlated better with
myocardial oxygen demand than EDP.
Barnea (41,42), Smith (43) and their co-workers in
1989 proposed a fitness function for optimal control of
deflation time. It included weighted MDP and PSP. The
permitted interval for deflation time was 200 ms to
100 ms relative to the predicted arrival of the next R
wave. An automatic controller was developed that used a
search and approximation algorithm to converge upon the
optimum fitness after a number of heartbeats. It was
tested successfully on computer simulations and anaesthetised dogs. It was able to follow a moving optimum,
both within the same patient over time and between
different patients.
Zelano et al. in 1990 (44) developed an automatic controller for the IABP that used different trigger signals.
Balloon inflation occurred either upon detection of S2 or at
a set time prior to the predicted time of the next S2. Balloon
deflation occurred either during the PR interval at a set
time after the P wave or after the R wave. The advantage of
using the P wave, R wave and S2 for triggering is that all
can be detected in real time. This allows the controller to
follow changes in heart rate and rhythm. It was tested
successfully in a semiautomatic open loop operation in
anaesthetised dogs with a coronary artery tied to simulate
a myocardial infarction. They proposed a fitness function
for automatic closed loop control. It used weighted MSP
and MDP.
Kantrowitz et al. in 1992 (45) reported a clinical trial of
automatic closed loop control of the IABP. They used a rulebased algorithm to adjust the time of inflation and deflation. Its safety was verified in anaesthetized dogs and it
was then tested on 10 human patients. Their aims were for
inflation to occur at dichrotic notch and for deflation to
overlap the first half of ventricular ejection. They were
successful in 99 and 100% of recordings respectively. Eight
of the patients survived. Neither of the two deaths was
attributed to the controller.
170
Sakamoto et al. in 1995 (46) developed a new algorithm

to cope with atrial fibrillation, the most unpredictable
cardiac arrhythmia sometimes described as irregularly
irregular. The aim was for inflation to occur at the dichrotic
notch. They were able to predict the time of arrival of the
dichrotic notch from mathematical analysis of the RR
interval from the previous 60 heartbeats. Deflation
occurred at the R wave. It was tested on ECG recordings
from real patients and performed better than conventional
timing.
FUTURE DEVELOPMENTS
The use of IABP therapy for preoperative support and as an
adjunct to coronary angioplasty or bypass and off-bypass
cardiac surgery is likely to increase although larger studies
are required to identify which patients will benefit most. The
role of the IABP in left ventricular failure is likely to decrease
with the increasing use of left ventricular assist devices.
Manufacturers recent research and development has
focused on: Improved automatic control algorithms that
better cope with alterations in heart rate and rhythm and
adjust inflation and deflation times to optimize cardiac
assistance. Better catheter designs that cause fewer vascular complications and permit more rapid movement of
the shuttle gas.
BIBLIOGRAPHY
1. Harken DE. Presentation at the International College of Cardiology, Brussels; 1958.
2. Clauss RH, et al. Assisted circulation. I. The arterial counterpulsator. J Thorac Cardiovasc Surg 1961;41:447.
3. Kantrowitz A, McKinnen WMP. Experimental use of diaphragm as experimental myocardium. Surg Forum 1958;9:266.
4. Moulopoulos SD, Topaz S, Kolff WJ. Diastolic balloon pumping
(with carbon dioxide) in the aorta. A mechanical assistance to
the failing circulation. Am Heart J 1962;63:669.
5. Clauss RH, Missier P, Reed GE, Tice D. Assisted circulation by
counterpulsation with intra-aortic balloon: Methods and
effects. Proceeding of the 4th ACEMB; 1962.
6. Kantrowitz A, et al. Initial clinical experience with intraaortic
balloon pumping in cardiogenic shock. J Am Med Ass
1968;203:113.
7. Dennis C, Moreno JR, Hall DP. Studies on external counterpulsation as a potential measure for acute left heart failure.
Trans Am Soc Artif Intern Organs 1963;9:186.
8. Kantrowitz A, et al. Initial clinical experience with a new
permanent mechanical auxiliary ventricle: The dynamic aortic
patch. Trans Am Soc Artif Intern Organs 1972;18(0):159167,
179.
9. Kantrowitz A, et al. Jr. Initial clinical experience with
intraaortic balloon pumping in cardiogenic shock. JAMA
1968;203(2):113118.
10. Kantrowitz A. Origins of intraaortic balloon pumping. Ann
Thoracic Surg 1990;50(4):672674.
11. Freed PS, Wasfie T, Zado B, Kantrowitz A. Intraaortic balloon
pumping for prolonged circulatory support. Am J Cardiol
1988;61(8):554557.
12. Lund O, et al. Intraaortic balloon pumping in the treatment of
low cardiac output following open heart surgeryimmediate
results and long-term prognosis. Thoracic Cardiovascular
Surg 1988;36(6):332337.
13. Meharwal ZS, Trehan N.Vascular complications of intra-aortic balloon insertion in patients undergoing coronary reavscularization: Analysis of 911 cases.[See comment]. Eur J CardioThoracic Surg 2002;21(4):741747.
14. Ferguson JJ. 3rd, et al. The current practice of intra-aortic
balloon counterpulsation: Results from the benchmark registry. J Am College Cardiol 2001;38(5):14561462.
15. Dietl CA, et al. Efficacy and cost-effectiveness of preoperative
iabp in patients with ejection fraction of 0.25 Or less. Ann
Thorac Sur 1996;62(2):401408.
16. Christenson JT, Simonet F, Badel P, Schmuziger M. Evaluation of preoperative intra-aortic balloon pump support in high
risk coronary patients. Eur J Cardio-Thoracic Sur 1997;11(6):
10971103.
17. Suzuki T, et al. Usefulness of preoperative intraaortic
balloon pump therapy during off-pump coronary artery bypass
grafting in high-risk patients. Ann Thoracic Surg 2004;77
(6):20562059.
18. Brodie BR, Stuckey TD, Hansen C, Muncy D. Intra-aortic
balloon counterpulsation before primary percutaneous
transluminal coronary angioplasty reduces catheterization
laboratory events in high-risk patients with acute myocardial
infarction. Am J Cardiol 1999;84(1):1823.
19. Ishihara M, Sato H, Tateishi H, Uchida T, Dote K. Intraaortic
balloon pumping as the postangioplasty strategy in
acute myocardial infarction. Am Heart J 1991;122(2):385
389.
20. Ohman EM, et al. Use of aortic counterpulsation to improve
sustained coronary artery patency during acute myocardial
infarction. Results of a randomized trial. The randomized iabp
study group Circulation 1994;90(2):792799.
21. Stone GW, et al. A prospective, randomized evaluation of
prophylactic intraaortic balloon counterpulsation in high risk
patients with acute myocardial infarction treated with primary angioplasty. Second primary angioplasty in myocardial
infarction (pami-ii) trial investigators. J Am College Cardiol
1997;29(7):14591467.
22. vant Hof AW, et al. A randomized comparison of intra-aortic
balloon pumping after primary coronary angioplasty in high
risk patients with acute myocardial infarction. [See comment].
Eur Heart J 1999;20(9):659665.
23. Sanborn TA, et al. Impact of thrombolysis, intra-aortic balloon
pump counterpulsation, and their combination in cardiogenic
shock complicating acute myocardial infarction: A report from
the shock trial registry. Should we emergently revascularize
occluded coronaries for cardiogenic shock? J Am College Cardiol 2000;36(3 Suppl. A):11231129.
24. Minich LL, et al. Intra-aortic balloon pumping in children with
dilated cardiomyopathy as a bridge to transplantation. J Heart
Lung Transplant 2001;20(7):750754.
25. Pantalos GM, et al. Estimation of timing errors for the
intraaortic balloon pump use in pediatric patients. ASAIO
Journal 1999;45(3):166671.
26. Akomea-Agyin C, et al. Intraaortic balloon pumping in children. Ann Thoracic Surg 1999;67(5):14151420.
27. Pinkney KA, et al. Current results with intraaortic balloon
pumping in infants and children. Ann Thoracic Surg
2002;73(3):887891.
28. Kantrowitz A, et al. Intraaortic balloon pumping 1967 through
1982: analysis of complications in 733 patients. Am J Cardiol
1986;57(11):976983.
29. Miller JS, Dodson TF, Salam AA, Smith RB3rd. Vascular
complications following intra-aortic balloon pump insertion.
Am Surg 1992;58(4):232238.
30. Macoviak J, et al. The intraaortic balloon pump: an analysis of
five years experience. Ann Thoracic Surg 1980;29(5):451
458.

31. Vonderheide RH, Thadhani R, Kuter DJ. Association of thrombocytopenia with the use of intra-aortic balloon pumps. Am J
Med 1998;105(1):2732.
32. Jaron D, Ohley W, Kuklinski W. Efficacy of counterpulsation:
model and experiment. Trans Am Soc Artificial Inter Organs
1979;25:372377.
33. Jaron D, Moore TW, He P. Theoretical considerations regarding the optimization of cardiac assistance by intraaortic balloon
pumping. IEEE Trans Biome Eng 1983;30(3):177185.
34. Jaron D, Moore TW, He P. Control of intraaortic balloon
pumping: theory and guidelines for clinical applications.
Ann Biomed Eng 1985;13(2):155175.
35. Niederer P, Schilt W. Experimental and theoretical modelling
of intra-aortic balloon pump operation. Med Biol Eng Comput
1988;26(2):167174.
36. Barnea O, Smith B, Moore TW, Jaron D. Simulation and
optimization of intra-aortic balloon pumping. Proceedings.
Computers in Cardiology (Cat. No.89CH2932-2). Los Alamitos
(CA): IEEE Computer Society Press; 1990. p 237240.
37. Sakamoto T, et al. Effects of timing on ventriculoarterial
coupling and mechanical efficiency during intraaortic balloon
pumping. ASAIO Journal 1995;41(3):M580MM583.
38. Weller PR, Morrow DR, LeFe`vre JE. Evolution of a fuzzy
controller for the intra-aortic balloon pump. Proceedings of
2nd European Medical and Biological Engineering Conference, EMBEC02, Vienna, Austria, Hutten H, Krosl P. editors.
ISBN 3-901351-62-0;48. December 2002; p 15881589.
39. Kane GR, Clark JW, Bourland HM, Hartley CJ. Automatic
control of intra-aortic balloon pumping. Trans Am Soc Artif Int
Organs 1971;17:148.
40. Martin PJ, Jaron D. New controller for in-series cardiac-assist
devices. Med Biol Eng Computing 1978;16(3):243249.
41. Barnea O, Smith B, Moore TW, Jaron D. An optimal control
algorithm for intra-aortic balloon pumping. Images of the
Twenty-First Century. Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and
Biology Society (Cat. No.89CH2770-6). New York: Vol. 5.
IEEE; 1989. p 14191420.
42. Barnea O, et al. Optimal controller for intraaortic
balloon pumping. IEEE Trans. Biomed Eng 1992;39(6):629
634.
43. Smith B, Barnea O, Moore TW, Jaron D. An algorithm for
optimal control of the intra-aortic balloon pump. Proceedings
of the Fifteenth Annual Northeast Bioengineering Conference
(Cat. No.89-CH2689-8). New York; IEEE. 1989; p 7576.
44. Zelano JA, Li JK, Welkowitz W. A closed-loop control scheme
for intraaortic balloon pumping. IEEE Trans Biomed Eng
1990;37(2):182192.
45. Kantrowitz A, et al. Initial clinical trial of a closed loop, fully
automatic intra-aortic balloon pump. ASAIO Journal 1992;
38(3):M617M621.
46. Sakamoto T, Arai H, Maruyama T, Suzuki A. New algorithm of
intra aortic balloon pumping in patients with atrial fibrillation. ASAIO Journal 1995;41(1):7983.
Further Reading
Goldberger M, Tabak SW, Shah PK. Clinical experience with
intra-aortic balloon counterpulsation in 112 consecutive
patients. Am Heart J 1986;111(3):497502.
McEnany MT, et al. Clinical experience with intraaortic balloon
pump support in 728 patients. Circulation 1978;58(3 Pt 2):I124
I132.
See also ARTERIES, ELASTIC PROPERTIES OF; HEMODYNAMICS; VASCULAR GRAFT
PROSTHESIS.
171
INTRACRANIAL PRESSURE MONITORING. See

MONITORING,
INTRACRANIAL PRESSURE.
INTRAOCULAR LENSES. See LENSES, INTRAOCULAR.

INTRAOPERATIVE RADIOTHERAPY. See
RADIOTHERAPY,
INTERAOPERATIVE.
INTRAUTERINE DEVICES (IUDS). See

CONTRACEPTIVE
DEVICES.

JOSEPH P. BRUNER
Department of Obstetrics and
Gynecology
Nashville, Tennessee
R. DOUGLAS WILSON
N. SCOTT ADZICK
University of Pennsylvania
Philadelphia, Pennsylvania
LOUISE WILKINS-HAUG
AUDREY C. MARSHALL
Womens and Childrens Hospital
Boston, Massachusetts
RUBEN A. QUINTERO
Florida Institute for Fetal
Diagnosis and Therapy
Tampa, Florida
M. YAMAMOTO
Y. VILLE
University of Paris
Paris, France
ANTHONY JOHNSON
University of North Carolina
Chapel Hill, North Carolina
JULIE S. MOLDENHAUER
Wayne State University
Detroit, Michigan
INTRODUCTION
Intrauterine surgery of the fetus or fetal adnexae is spreading rapidly throughout the world. In a broad sense, intrauterine surgery includes any procedure in which a medical
device is purposely placed within the uterine cavity. For
most physicians, however, the concept of intrauterine surgery excludes such commonly performed procedures as
amniocentesis and chorionic villous sampling. Rather,
the term usually refers to techniques requiring specialized
knowledge, experience, and most especially, instrumentation. Procedures most commonly mentioned in discourses
on intrauterine surgery include those specifically developed for the treatment of such serious anatomic defects
as congenital cystic adenomatoid malformation (CCAM),
sacrococcygeal teratoma (SCT), lower urinary tract
obstruction (LUTO), myelomeningocele, aortic or pulmonic
stenosis, gastroschisis, iatrogenic amniorhexis, twin-totwin transfusion syndrome (TTTS), and twins discordant
for severe anomalies. As no one person in the world is an
expert in every one of these procedures, the remainder of
172
this chapter is divided into individual sections, each

authored by someone widely recognized as a leader in
the treatment of that particular anomaly.
CONGENITAL CYSTIC ADENOMATOID MALFORMATION

AND SACROCOCCYGEAL TERATOMA
Most prenatally diagnosed malformations are managed by
appropriate medical and surgical evaluation and treatment following planned delivery near term, with some
cases requiring transfer of the mother to a tertiary referral
center with obstetrics, maternal fetal medicine, medical
genetics, neonatology, and pediatric surgery subspecialties.
Certain anatomic abnormalities can have significant fetal
developmental consequences, with emergency in utero therapy being required due to gestational age and mortality
risks for the fetus. Open maternal fetal surgery poses additional risks to the mother. Maternal fetal surgery (14)
should not be attempted until (1) the natural history of
the fetal disease is established by following up on untreated
cases, (2) selection criteria for cases requiring intervention
are developed, (3) pathophysiology of the fetal disorder and
its correction are defined in fetal animal models, and (4)
hysterotomy and fetal surgery can be performed without
undue risk to the mother and her reproductive potential.
Congenital cystic adenomatoid malformation of the lung
(CCAM) (5) is a rare lesion characterized by a multicystic
mass of pulmonary tissue with proliferation of bronchial
structures. CCAM is slightly more common in males and is
unilobar in 8095% of cases. CCAMs derive their arterial
blood supply from the normal pulmonary circulation.
CCAMs are divided clinically into cystic and solid lesions,
but have been divided traditionally into three types based
on their pathological characteristics. Type 1 CCAM lesions
account for 50% of postnatal CCAM cases and consist of
single or multiple cysts lined by ciliated pseudostratified
epithelium (5). These cysts are usually 310 cm in size and
14 in number (5). Type II CCAM lesions account for 40% of
postnatal cases of CCAM and consist of more numerous
cysts of smaller diameter, usually less than 1 cm. They are
lined by ciliated, cuboidal, or columnar epithelium (5).
Type III CCAM lesions account for only 10% of CCAM
cases and are usually large, homogenous, microcystic
masses that cause mediastinal shift. These lesions have
bronchiolar-like structures lined by ciliated cuboidal
epithelium separated by masses of alveolar-sized structures lined by nonciliated cuboidal epithelium (5). Prognosis in Type III CCAM is related to size.
Sacrococcygeal teratoma (SCT) (6) is defined as a neoplasm composed of tissues from either all three germ layers
or multiple foreign tissues lacking an organ specificity.
SCT is thought to develop from a totipotent somatic cell
originating in Hensens node. SCT has been classified by
the relative amounts of presacral and external tumor present, with Type I completely external with no presacral
component, Type II external component with internal
pelvic component, Type III external component with internal component extending into the abdomen, and Type IV
completely internal with no external component (7). A Type
I SCT is evident at birth and is usually easily resected and
has a low malignant potential (6). Type II and III SCTs are
recognized at birth, but resection may be difficult requiring
an anterior and posterior approach (6). A Type IV SCT can
have a delayed diagnosis until symptomatic at a later age
(6). SCT is one of the most common tumors in newborns and
has an incidence of 1 per 35,000 to 40,000 live births (6).
Evaluation of the fetal status for CCAM and SCT
requires multiple imaging and functional techniques
(810), including fetal ultrasound, fetal MRI, and fetal
echocardiogram with arterial and venous Doppler assessments (umbilical artery, umbilical vein, ductus venosus).
Measurements include combined cardiac output, cardiothoracic ratio, descending aortic blood flow, inferior vena
cava diameter, placental thickness, umbilical artery systolic to diastolic Doppler ratio, and amniotic fluid index.
Presence of ascites, pleural or pericardial effusion, and
skin or scalp edema are important markers for the extent
of fetal hydrops and its overall effect on fetal stability.
Specific ultrasound imaging of the CCAM and SCT looks
for the percentage of cystic and solid components in the
tumors as well as an overall mass volume (cc) estimate (AP
cm transverse cm height cm 0.52). The SCT consistency and size can be reflected directly in the combined
cardiac output and amount of vascular shunting. The
CCAM overall size can cause mediastinal shift with cardiac
dysfunction and pulmonary deformation. Validated ratio of
CCAM/head circumference (CVR) can be used for prognosis
and follow-up planning (10). The specific lobar location for
the CCAM may have a differential impact on cardiac
function. The development of fetal hydrops is due mainly
to cardiac dysfunction secondary to compression.
The physiologic changes required in the fetal status to
move from expectant management to open maternal fetal
surgery is generally dictated by fetal (gestational age and
extent of fetal hydrops) and maternal factors (810). Criteria for consideration of maternal fetal surgery for CCAM
resection (fetal lobectomy) require the absence of maternal
risk factors for anesthesia and surgery, a singleton pregnancy with a normal karyotype (amniocentesis, chorionic
villus sampling, or percutaneous umbilical blood sampling),
no other anatomical abnormalities beyond the associated
hydrops, gestational age of 2131 weeks, and massive multicystic or predominantly solid CCAM (CVR > 1.6) (810). In
selected cases, the failure of in utero therapy techniques,
such as thoracoamniotic shunting or cyst aspiration for the
large Type I lesions, to reverse the fetal hydrops would be
required. Criteria for consideration of maternal fetal surgery for debulking of a SCT require the absence of maternal
risk factors for anesthesia and surgery, a singleton pregnancy with a normal karyotype, the absence of significant
associated anomalies, evidence of impending high output
cardiac failure, gestational age of 2130 weeks, and favorable SCT anatomy classification (Type I or II) (9).
The technique for maternal hysterotomy to allow access
to the fetus has been well described and has evolved over 25
years of experimental and clinical work (1,8,9). The uterus
is exposed through a maternal low transverse abdominal
incision. If a posterior placenta is present, superior and
inferior subcutaneous flaps are raised and a vertical midline fascial incision is made to expose the uterus for a
convenient anterior hysterotomy with the uterus remaining
in the abdomen. Conversely, the presence of an anterior

placenta necessitates the division of the rectus muscles so
the uterus can be tilted out of the abdomen for a posterior
hysterotomy. A large abdominal ring retractor (Turner
Warwick) is used to maintain exposure and prevent lateral
compression of the uterine vessels. Sterile interoperative
ultrasound is used to delineate the fetal position and placental location. The edge of the placenta is marked under
sonographic guidance using electrocautery or a marking
pen. The position and orientation of the hysterotomy is
planned to stay parallel to and at least 6 cm from the
placental edge but still allow exposure of the appropriate
fetal anatomy. The hysterotomy is facilitated by the placement of two large monofilament sutures (PDS II 1 Ethicon;
Somerville, NJ) parallel to the intended incision site and
through the full thickness of the uterine wall and membranes under sonographic guidance. The electrocautery is
used to incise the myometrium between the two stay sutures
down to the level of the amniotic membranes. A uterine
stapler device (US Surgical Corporation; Norwalk, CT) with
absorbable Lactomer staples is then directly introduced
through the point of fixation and into the amniotic cavity
by using a piercing attachment on the lower limb of the
stapler. The stapler is fired, thereby anchoring the
amniotic membranes (chorion, amnion) to the uterine wall
creating a hemostatic hysterotomy. Careful evaluation for
the membrane adhesion status and for any myometrial
bleeding sites is undertaken. If required, interrupted PDS
sutures are used to control bleeding and membrane
separation. The fetus and the internal uterine cavity
are continually bathed in warmed lactated Ringers at
38408C using a level I warming pump connected to a
red rubber catheter that is placed in the uterine cavity
through the hysterotomy.
For CCAM resection (1,8,11), once the appropriate fetal
area is visualized in the hysterotomy site, the fetal arm is
brought out for pulse oximeter monitoring, IV access, and
fetal position control. Intraoperative fetal echocardiography is used throughout to monitor cardiac function. The
fetal chest is entered by a fifth intercostal space thoracotomy. The lesion usually decompresses out through the
thoracotomy wound consistent with the increase in the
thoracic pressure from the mass (8). Using techniques
initially developed on experimental animals, the appropriate pulmonary lobes containing the lesion are resected
(1,11). Fetal resuscitation is performed if needed through
intravenous administration of crystalloid, blood, and codeblue medications with fetal echocardiography providing
functional information. The fetal thoracotomy is closed
and the fetal arm is returned to the uterus.
The technique for debulking of an external fetal SCT has
been described in detail previously (1,9,12,13). The fetal
foot is used for pulse oximeter monitoring and IV access
with intraoperative echocardiography. The fetal SCT is
exposed and a Hagar dilator is placed in the rectum. Fetal
skin is incised circumferentially around the base of the
tumor and a tourniquet is applied to constrict blood flow.
The tumor is debulked externally, usually with a 90 mm
thick tissue stapler (US Surgical Corporation; Norwalk,
CT). The objective of the fetal SCT resection is to occlude
the tumor vascular supply and remove the low resistance
173
tumor vascular bed from the fetal circulation. No attempt is

made to dissect the intrapelvic component of the tumor or
to remove the coccyx (done with a second procedure after
birth). Fetal resuscitation is performed if needed through
intravenous administration of crystalloid, blood, and codeblue medications with fetal echocardiography providing
functional information. The fetal sacral wound is closed.
Repair of the hysterotomy after fetal surgery (14) uses
a water-tight two-layered uterine closure, with interrupted
full thickness stay sutures placed first and untied using
PDS II 1 (Ethicon; Somerville, NJ), and the uterus is then
closed with a running continuous stitch PDSII 0 (Ethicon;
Somerville, NJ) including the chorion-amnion membrane
layer. The interrupted stay sutures are then tied after the
amniotic fluid volume has been corrected with warm lactated Ringers through a red rubber catheter and volume
confirmed by ultrasound visualization. The omentum is
sutured in place over the hysterotomy closure to help seal
the hysterotomy site with vascularized tissue and to prevent bowel adherence to the site, especially when a posterior hysterotomy is performed. The maternal laparotomy
incision is closed in layers. It is important to use a subcuticular skin closure covered with a transparent dressing
so that monitoring devices can be placed on the maternal
abdomen postoperatively.
In some specific cases, when the CCAM lesion is not
resected in utero, it continues to be a large space-occupying
lesion with mediastinal shift. Thus, it might be anticipated
that respiratory compromise will be present at birth, the
delivery may be facilitated with an EXIT procedure
(ex utero intrapartum therapy) (14). Uterine relaxation
is maintained by high concentration inhalational anesthetics, with additional tocolysis if necessary. The EXIT
requires only the head and chest to be initially delivered
through, preferably, a low transverse hysterotomy wound
thereby preserving uterine volume with the lower fetal
body and continuous warmed lactated Ringers infusion to
prevent cord compression. These maneuvers preserve the
uterine-placental circulation and continue placental gas
exchange. The EXIT procedure can be done through an
anterior or posterior hysterotomy, but its location in the
uterus may require that all future pregnancies be delivered
by cesarean section with no trial of labor if a low anterior
transverse location is not available.
All future pregnancies following maternal hysterotomy
for maternal-fetal surgery require cesarean section at term
with no trial of labor. Maternal obstetrical risks in a
subsequent pregnancy are similar to risks following for a
classic cesarean section (15).
LOWER URINARY TRACT OBSTRUCTION

The diagnosis and treatment of fetal lower urinary tract
obstruction (LUTO) requires knowledge of the differential
diagnosis and the natural history of the condition, a thorough understanding of the criteria for therapy, and management expertise. Fetal LUTO is one of the most
commonly diagnosed birth defects. Untreated, and depending on the level of the obstruction, it may lead to hydronephrosis, renal dysplasia, pulmonary hypoplasia, and
174
perinatal death (16,17). The prognosis depends on the

extent of preexisting renal damage and the effectiveness
of therapy. Treatment with fetal urinary diversion procedures is aimed at preventing renal damage and pulmonary
hypoplasia (1820).
Obstruction to urine flow has been shown in animal
models to result in hydronephrosis and renal dysplasia
(21). Release of the obstruction is associated with no or
variable renal damage depending on the timing of the
release or the creation of the defect (21,22). Pulmonary
hypoplasia is another major potential complication of
fetuses with obstructive uropathy (23). The association
probably results from the attendant oligohydramnios.
Urethral obstruction may result from posterior urethral
valves (PUV), anterior urethral valves, megalourethra,
urethral duplications, urethral atresia, obstructive ureterocele, or cloacal dysgenesis. Posterior urethral valves
(PUV), first described by Young et al. (24), constitute the
most common cause of lower urinary tract obstruction in
male neonates, with an incidence of 1:8000 to 1:25,000
livebirths (25). The lesions occur only in males because
the female counterpart of the verumontanum, from which
the valves originate, is the hymen.
In utero therapy is usually limited to fetuses with
bladder outlet obstruction. Fetuses with unilateral
obstruction are not typically considered candidates for in
utero therapy, regardless of the magnitude of the obstruction or renal findings. In these patients, the risk/benefit
ratio of in utero intervention favors expectant management, even if it means loss of the affected renal unit.
Fetal renal function may be assessed by analysis of fetal
urinary parameters via vesicocentesis. Patients are considered candidates for in utero therapy if fetal urinary
parameters are below the threshold for renal cystic dysplasia. If the values are above the threshold, therapy
should not be offered.
The application of selection criteria in patients with
fetal LUTO for possible in utero therapy results in a significant attrition rate. Disqualification from therapy may
result both from too healthy or too sick conditions.
Examples of too healthy conditions include normal amniotic fluid volume or suggestion of nonobstructive dilatation
of the urinary tract. Examples of too sick conditions include
sonographic evidence of renal cystic dysplasia, abnormal
fetal urinary parameters, abnormal karyotype, or the presence of associated major congenital anomalies. Of 90
patients referred to the Florida Institute for Fetal Diagnosis and Therapy from October 1996 to October 2003,
more than one-half were disqualified from therapy from
single or overlapping conditions.
Percutaneous ultrasound-guided vesicoamniotic shunting of fetuses with LUTO began in the early 1980s
(16,19,23). The goal of therapy is to avoid development
of pulmonary hypoplasia from the attendant oligohydramnios as well as to preserve renal function. Fetal bladder
shunting should be offered only to patients without sonographic or biochemical evidence consistent with renal cystic dysplasia, normal karyotype, and lack of associated
major congenital anomalies.
The procedure can be performed under local, regional,
or general anesthesia. A minimal skin incision is made.
Ultrasound is used to identify the ideal site of entry into the

fetal bladder, below the level of the umbilicus. Color Doppler ultrasonography is used to identify the umbilical
vessels around the distended bladder and avoid them.
Under ultrasound guidance, the trocar is directed through
the maternal tissues and up to the fetal skin. Fetal analgesia is achieved with pancuronium 0.2 mg/kg and fentanyl
10 mcg/kg. The trocar stylet is used to enter the fetal
bladder with a sharp, swift, and controlled maneuver. If
a prior vesicocentesis had been performed, it is advisable to
obtain a sample of fetal urine for microbiological purposes
to rule out preexisting infection. A sample of fetal urine is
sent for further biochemical testing. Placement of the
double-pigtail catheter is monitored with ultrasound. After
the distal loop is deployed in the bladder, the trocar is
retrieved to the level of the bladder wall. A small amount of
the straight portion of the catheter may be advanced into
the bladder to avoid retracting the distal loop into the
bladder wall. The trocar shaft is retrieved slowly while
simultaneously maintaining pressure on the catheter to
deploy the straight portion within the bladder wall and
fetal skin. Once the shaft of the trocar reaches the fetal
skin, entrance of the catheter, including the proximal loop,
can be safely deployed. If complete anhydramnios is present prior to insertion of the catheter, it is advantageous to
attempt an amnioinfusion with an 18 gauge needle prior to
shunting to create the space for deployment of the proximal
loop. Amnioinfusion is aimed at preventing misplacement
of the proximal loop within the myometrium and fetal
membranes.
Despite adequate placement, malfunction of vesicoamniotic shunting may occur up to 60% of the time (26). The
shunt may pull from the skin into the fetal abdomen,
resulting in iatrogenic ascites, or out of the fetal bladder,
with no further drainage of urine. The shunt may pull out
of the fetus altogether as well. Replacement of the shunt is
associated with an additive risk of fetal demise, chorioamnionitis, premature rupture of membranes, and miscarriage or preterm delivery, for a total perinatal loss rate of
approximately 5% per instance.
In 1995, we proposed the use of endoscopy to assess the
fetal bladder for diagnostic and surgical purposes (27,28).
Endoscopic visualization of the fetal bladder with a larger
endoscope can be justified during vesicoamniotic shunting.
Currently, we use a 3 mm or a 3.9 mm trocar with a
2.7 mm or 3.3 mm diagnostic or operating endoscope. This
diameter is slightly larger than the 14 gauge (approximately 2.1 mm) needle used for the insertion of the double-pigtail catheter. Access to the fetal bladder allows
remarkable evaluation of the bladder, ureteral orifices,
and urethra as well as the opportunity to perform surgical
procedures.
In normal fetuses, the urethra is not dilated, appearing
as a small hole within the bladder. In patients with a true
urethral obstruction, endoscopy will show a variable dilatation of the urethra at the level of the bladder neck. The
urethra is located using a 258 or a 708 diagnostic rigid
endoscope. Alternatively, a flexible/steerable endoscope
may be used. The anatomical landmarks to identify at this
level include the verumontanum and the urethral valves.
The diagnostic endoscope is then exchanged for a rigid
operating endoscope. A 600 mm YAG-laser fiber is passed

through the operating channel of the endoscope, and then
ablated using 510 w and 0.2 s pulses in successive steps.
The fiber is placed as anterior and medial as possible. It is
not necessary to evaporate the entire valvular tissue.
Instead, only a few defects to either side of the midline
are necessary to establish urethral patency (27,29). The
dilated urethra may collapse intraoperatively once patency
is re-established, which may obscure the field of view and
require frequent instillation of saline to the side port of the
trocar to distend it. Color Doppler may also be used to
document fetal urination through the penis.
A urethrorectal fistula may occur from thermal
damage beyond the posterior wall of the urethra into
the perirectal space. To avoid this complication, only 5
10 w of energy in short bursts should be used while ablating the valves.
The management of fetuses with lower obstructive
uropathy continues to be one of the most challenging
subjects in fetal therapy. The difficulties include establishing the correct differential diagnosis, accurately predicting subsequent renal function, and providing the best
treatment.
MYELOMENINGOCELE
Myelomeningocele results from the failure of caudal neural
tube closure during the fourth week of gestation. The lesion
is characterized by protrusion of the meninges through a
midline bony defect of the spine, forming a sac containing
cerebrospinal fluid and dysplastic neural tissue. Affected
infants exhibit varying degrees of somatosensory loss,
neurogenic sphincter dysfunction, paresis, and skeletal
deformities (30). Virtually all such infants also have the
Chiari II malformation, and up to 95% develop hydrocephalus (31). Although myelomeningocele is not a lethal
disorder, the neurologic sequelae are progressive, and
worsen until the lesion is closed. Observational and cohort
studies have demonstrated improvement of the Chiari II
malformation (32,33), decreased hydrocephalus (34), and
improved lower extremity function after intrauterine
repair of myelomeningocele (35).
On the day of surgery, the pregnant patient is taken to a
standard obstetrical operating room. An epidural catheter
is placed and, after induction of general endotracheal
anesthesia, she is prepared as if for a cesarean section.
Many of the general anesthetic agents cross the placenta
and provide analgesia for the fetus, and the epidural
catheter enables the administration of continuous postoperative analgesics if needed. The gravid uterus is
exposed with a Pfannenstiel incision and exteriorized.
The uterine contents are then mapped with a sterile ultrasound transducer, and the location of the fetus and the
placenta are determined. Initial uterine entry is obtained
with a specialized trocar developed at Vanderbilt University Medical Center (Cook Incorporated; Bloomington, IN).
The TulipanBruner trocar consists of a tapered central
introducer covered by a peel-away Teflon sheath. Use of
this trocar has demonstrated to reduce operative time and
blood loss while providing atraumatic entry into the uter-
175
ine cavity (36). Two through-and-through chromic sutures

are passed through the uterine wall and membranes on
either side of the selected entry point. The introducer is
then passed into the uterine cavity under direct ultrasonographic guidance using a modified Seldinger technique.
The central introducer is then removed, leaving only the
trocar sheath. Excess amniotic fluid may be aspirated and
stored in sterile, warm syringes. The footplate of a U.S.
surgical CS-57 autostapling device (United States Surgical
Corporation; Norwalk, CT) is then inserted through the
peel-away sheath, and the sheath is removed, leaving the
stapler in proper position. When activated, the stapler
creates a 68 cm uterine incision. At the same time, all
the layers of the uterine wall are held together, much like
the binding of a book.
The fetus is directly visualized and manually positioned
within the uterus so that the myelomeningocele sac is
located in the center of the hysterotomy. Proper position
is maintained by grasping the fetal head and trunk
through the flaccid uterine wall. During the procedure,
the fetal heart rate is monitored by continuous ultrasonographic visualization.
The myelomeningocele is closed in routine neurosurgical
fashion. Approximately 20% of patients will not have a wellformed myelomeningocele sac, but a crater-like lesion
termed myeloschisis. As fetuses with myeloschisis have less
viable skin for closure, it may be necessary to use bilateral
vertical relaxing incisions in the flanks to create bipedicular
flaps that can be advanced and closed over the dural sac. The
resulting full-thickness cutaneous defects are covered with
cadaveric skin (37).
After repair of the spina bifida lesion, the uterus is closed
in layers using #1 PDS sutures. The first layer incorporates
the absorbable polyglycolic acid staples left by the autostapling device. As the last stitches of this layer are placed, the
reserved amniotic fluid or physiologic crystalloid solution,
mixed with 500 mg of nafcillin or an equivalent dosage of an
antibiotic effective against Staphylococcus species, is
replaced in the uterus. The sterile, warm fluid is added
until the uterine turgor, as determined by manual palpation, is restored to the preoperative level, which is followed
by an imbricating layer. A sheet of Interceed absorbable
adhesion barrier (Johnson & Johnson Medical, Inc.; Arlington, TX) or omentum is attached over the incision to prevent
adhesion formation. The uterus is returned to the abdomen.
The fascial layer is closed in routine fashion, and the dermis
closed with a running subcuticular suture or staples. The
fetus is monitored postoperatively using continuous electronic fetal monitoring (EFM) and intermittent transabdominal ultrasonography.
Postoperative uterine contractions are monitored using
continuous EFM. Uterine contractions are initially controlled with intravenous magnesium sulfate and oral or
rectal indomethacin, and subsequently with subcutaneous
terbutaline or oral nifedipine, supplemented by indomethacin as needed. Patients are monitored with weekly
transabdominal ultrasonographic examinations. Delivery
of each child is accomplished via standard cesarean section.
Although the same abdominal incision is used for the
cesarean section as for the fetal surgery, the fetus is
preferably delivered via a lower uterine segment incision.
176
The uterus and abdominal incisions are closed in routine

fashion.
VALVULOPLASTY
Severe aortic stenosis in midgestation may lead to left
ventricular myocardial damage and can ultimately result
in hypoplastic left heart syndrome. Paradoxically, in these
fetuses, which are likely to progress to HLHS, the left
ventricle initially appears normal in size, or even enlarged,
in the setting of left ventricular systolic dysfunction. As
gestation progresses, diminished flow through the diseased
left ventricle leads to decreased flow, and the ventricle
experiences growth arrest, resulting in left heart hypoplasia at birth. Early relief of fetal aortic stenosis may preserve left heart function and growth potential by
maintaining flow through the developing chamber. To this
end, a number of operators have developed techniques to
perform fetal aortic valvuloplasty in second trimester
fetuses.
The mother is placed under general anesthesia in a
supine position with left lateral uterine displacement.
Transabdominal ultrasound imaging and external manipulation are employed to achieve ideal fetal position. In this
position, a line of approach from the anterior abdominal
surface traverses the apex of the fetal left ventricle (LV),
paralleling the LV outflow tract, and crossing the valve into
the ascending aorta. The fetus is given intramuscular
anesthetic and muscle relaxant prior to catheterization.
If unable to position the fetus using external maneuvers,
the operators perform a limited laparotomy to enable direct
uterine manipulation and transuterine imaging.
A low profile, over-the-wire coronary angioplasty catheter is chosen with a balloon diameter based on the measurement of the aortic annulus, using a balloon:annulus
ratio of 1:2. The balloon catheter is mounted on a floppytipped guidewire, with 3 cm of distal wire exposed. The
wire/catheter assembly is then advanced through the 19G
12 cm stainless-steel introducer cannula until the balloon
emerges. Affixing a visible and palpable marker on the
proximal catheter shaft allows the operator to reproduce
this balloon/cannula relationship during the procedure
without relying wholly on the ultrasound imaging.
The introducer is advanced through the fetal chest wall
and to the LV epicardium under ultrasound guidance. The
LV is entered with the introducer, and the obturator
removed with the tip of the cannula just below the aortic
valve (Fig. 1). Blood return through the cannula confirms
an intracavitary position.
The wire/catheter assembly is passed through the cannula, and the tip of the wire is identified as it emerges.
While maintaining imaging of the aortic valve and ascending aorta, the precurved wire tip is manipulated to probe
for the valve. Valve passage, confirmed echocardiographically by imaging the wire in the ascending aorta, is followed
by catheter insertion to the premarked depth. The balloon
is then inflated, by hand or by pressure gauge, to a pressure
at which it achieves the intended balloon:annulus ratio.
Upon completion of the dilation, the entire apparatus is
removed from the fetus.
Figure 1. Transabdominal ultrasound imaging during introducer

cannula insertion into dilated fetal left ventricle. The tip of the
introducer is positioned in the left ventricular outflow tract
directed at the stenotic aortic valve.
When first reported in the 1990s, fetal aortic valve

dilation was performed with minimal technical success
(38). Using the technique described above, technical success rates are now over 80% in fetuses between 21 and 26
weeks (39). As the technical aspects of the procedure continue to be refined, and safety is established, issues of
patient selection will become the major focus of ongoing
research. Anatomic and physiologic variables predicting
left ventricular normalization following successful fetal
aortic valvuloplasty remain poorly understood.
AMNIOEXCHANGE
Gastroschisis is a paraumbilical defect of the anterior
abdominal wall associated with intrauterine evisceration
of the fetal abdominal organs. The incidence of gastroschisis
is approximately 1:4000 births, with a 1:1 male:female
ratio. Most cases are sporadic and aneuploidy is uncommon.
Gastroschisis is characterized by a full-thickness defect
of the abdominal wall, usually located to the right of the
umbilical cord, which has a normal insertion. The defect in
the abdominal wall is generally quite small (35 cm). The
herniated organs include mainly bowel loops, although, in
rare cases, the spleen and liver may be involved. Intestinal
atresias and other gastrointestinal disruptions are found
in as many as 15% of cases, and malrotation is also universal.
Although the prognosis is excellent with an ultimate
survival of greater than 90%, many factors may jeopardize
the outcome of these infants. A chronic aseptic amniotic
fluid peritonitis (perivisceritis) often occurs. The herniated
organs become covered by an inflammatory peel in the
third trimester, resulting from chemical irritation by exposure to digestive enzymes in the amniotic fluid. Thickening, edema, and matting together of the intestines occurs in
these cases, and may result in a secondary ischemic injury
to the bowel as the abdominal defect becomes too small.

Meconium is frequently found in the amniotic fluid of
affected fetuses. Its presence probably reflects intestinal
irritation. Intrauterine growth restriction (IUGR) is frequent, occurring in up to 60% of fetuses. Oligohydramnios
can occur, and may lead to fetal stress by cord compression.
Premature birth is a frequent and still poorly understood
complication. At birth, infants have low serum albumin
and total protein levels, which probably results from
chronic peritonitis.
The obstetrical management of the fetus with gastroschisis is controversial. Some studies have shown no clear
benefit of cesarean delivery over vaginal delivery, where as
others demonstrate an improved perinatal outcome in
infants delivered by elective cesarean section prior to labor.
Postoperative infection and delayed total enteral nutrition
are the major acute complications of the newborn.
Although all neonates with gastroschisis require surgery
shortly after birth, repair may be by primary fascial closure, or by delayed fascial closure using temporary coverage with a silastic/Dacron intra-abdominal pouch. The
repair as a primary or secondary procedure depends on
the degree of chemical peritonitis with matting of the bowel
that is present. Delayed intestinal function with poor
enteral nutrition is expected in most patients. Central
venous access and early total parenteral nutrition are
therefore usually required.
In a study by Luton et al. (40), gastroschisis was created
at mid gestation in 21 lamb fetuses. Saline was amnioinfused in some fetuses every 10 days until term. Thickness
of the bowel muscularis, thickness of the serous fibrosis,
and plasma cell infiltration were all significantly improved
in amnioexchanged animals when compared with fetal
lambs that were not amnioexchanged (40). Histologic analysis of appendices removed from human newborns demonstrated increased fibrosis in those with gastroschisis; after
amnioinfusion, the serosa was still edematous, but no
inflammation was seen (41). In a pilot study in human
pregnancies (42), the same authors investigated the effect
of amnioinfusion on the outcome of prenatally diagnosed
gastroschisis. Following up on their work showing that an
inflammatory response exists in the amniotic fluid of
fetuses with gastroschisis, they hypothesized that amniotic
fluid exchange would improve the outcomes of prenatally
diagnosed cases. The outcome of 10 amnioinfused fetuses
with gastroschisis was compared with 10 nonamnioinfused
matched controls. Results showed that fetuses undergoing
amnioinfusion had a shorter duration of curarization after
surgical repair (2.2
1.9 versus 6.8
6.9 days,
0.019), a shorter delay before full oral feeding
(49.7
21.5 versus 72.3
56.6 days, NS) and a shorter
overall length of hospitalization (59.5
19.7 versus
88.5
73.6 days, NS). The authors confirmed their previous data showing that amniotic fluid displays a chronic
inflammatory profile, and they speculated that a reduction
of the inflammatory response could improve the outcome of
human fetuses with gastroschisis (42).
Amnioexchange for treatment of gastroschisis begins
after 30 weeks gestation, and is repeated approximately
every two weeks until delivery. A complete obstetrical
ultrasound examination is performed prior to the amnioex-
177
Anterior
Up
Down Right
Left
d
r
p
p
Posterior
Longitudinal view of the uterus
Transversal view of the uterus
Figure 2. Views of the uterus.
change. The patient is admitted to the labor and delivery

suite, and a nonstress test is performed before and after the
amnioexchange. An intravenous line or a heplock is placed,
and a single vial of blood is obtained and held for routine
admission laboratory studies, in the case that urgent
delivery is required. Prophylactic tocolysis may be given
in the form of intravenous or subcutaneous terbutaline.
Light IV sedation may also be given if desired.
After performing a sterile abdominal prep and drape,
amnioexchange is performed using a closed system. The
closed tubing system materials are illustrated in Fig. 2. All
of the materials are sterile except the graduated cylinder.
Two sterile three-way stopcocks (b and c) are connected
end-to-end. Sterile IV solution (a) is connected to stopcock
(c) by way of sterile IV tubing (a). Three lengths of sterile
connecting IV tubing (b) are connected to the remaining
exposed ports of the stopcock assembly. Tubing from the
side port of stopcock (b) is allowed to drain to the graduated
cylinder (d). A 60 mL syringe (e) is connected to the tubing
attached to the inline port of stopcock (b). The tubing
attached to the inline port of stopcock (c) will connect to
the therapeutic amniocentesis needle. Stopcock (c) is closed
off to the IV solution (a). Stopcock (b) is closed off to the
graduate (d). Amniocentesis is performed under continuous ultrasound guidance. Once access is obtained, the
stylet is removed from within the needle lumen and the
connection tubing is attached. With the stopcocks positioned as noted above, amniotic fluid is withdrawn into
the syringe (e) until the syringe is filled. Stopcock (b) is
then closed off to the patient and the fluid is expelled into
the graduate (d). Stopcock (b) is then turned off to the
graduate (d), and this step is repeated until the desired
amount of amniotic fluid has been withdrawn (300900
ml). With stopcock (b) closed to the graduate (d), stopcock
(c) is closed off to the patient. The syringe is then filled with
sterile warmed saline. Stopcock (c) is then closed to the IV
solution (a) and the fluid is infused into the patient. This
step is repeated until the desired amount of fluid is infused.
These steps are repeated serially until the infusion procedure is complete. If the amniotic fluid volume falls within
normal range, the amniotic fluid volume at the end of the
amnioexchange should be the same as at the beginning of
the procedure. In the presence of oligohydramnios, additional sterile warmed fluid can be added to the uterine
cavity in order to achieve a normal fluid volume by the end
of the procedure.
178
If the fluid volume is normal, the placenta is located

posteriorly or fundally, and the fetus is quiescent, all of the
toxic amniotic fluid planned for removal might be aspirated
in one step. With an anteriorly implanted placenta, however, or in the presence of oligohydramnios or an active
fetus, it may be necessary to remove a small amount of
amniotic fluid, and then replace it with warmed normal
saline, repeating the procedure serially until the amnioexchange is completed.
After completion of the amnioexchange, electronic monitoring of the fetal heart rate and uterine activity continues
until the patient fulfills the usual criteria for discharge.
the amniotic cavities through the membrane defect, causing dissection of the chorionic cavity. In these patients,
only the chorion separates this fluid from leaking grossly to
the vagina. In this group, the amniopatch was successful in
resealing the amniotic membrane in 7 of the 11 patients
(63.6%).
The precise mechanism by which the amniopatch works
is unknown. Presumably, platelet activation at the site of
rupture and fibrin formation initiates a healing process
that enables the membranes to seal.
AMNIOPATCH
Feto-fetal transfusion syndrome can be described in all

monochorionic multiple pregnancies but has been extensively reported in twins. Twin-to-twin transfusion syndrome (TTTS) develops in approximately 15% of all
monochorionic pregnancies (50), and carries a high perinatal mortality rate (50). The fetuses are morphologically
normal, and inter-twin vascular communications on the
chorionic plate are thought to be responsible for the
development of the disease through unidirectional blood
transfusion from the donor to the recipient twin. Besides
the primary hemodynamic imbalance between the twins,
the disease may lead to disruptive lesions in both twins.
Before the development of antenatal ultrasound, TTTS
was diagnosed at birth as a discordance of at least 20% in
weight and 5 g/dL in the hemoglobin concentrations of two
twins of the same sex (52). These criteria were abandoned
because these features could not be consistently recognized in utero. With the development of ultrasound, the
polyhydramnios-oligohydramnios sequence has been
found to be the condition carrying one of the highest
perinatal mortality rates in obstetrics, up to 90% without
treatment.
Laser coagulation of placental anastomoses by fetoscopy
is the most effective first-line treatment of FFTS, which
leads to at least one survivor at birth and intact survival at
6 months of age in 76% and 76% respectively, as compared
with 56% and 51% in cases treated by serial amnioreduction in the Eurofetus randomized trial (53).
The selection criteria to qualify for percutaneous endoscopy-directed laser coagulation of placental anastomoses
include:
Iatrogenic preterm premature rupture of membranes

(PPROM) occurs in approximately 1.2% of patients after
genetic amniocentesis (43), 35% of patients after diagnostic fetoscopy (44), and approximately 58% of patients after
operative fetoscopy. Although the membranes might seal
spontaneously in this setting (45,46), most patients continue to leak fluid and are at a risk for pregnancy loss.
The overall perinatal mortality of previable PPROM
managed expectantly is 60% (47,48). Nearly one-third of
these deaths occurs in utero. Pulmonary hypoplasia occurs
in 50% of cases diagnosed before 19 weeks (49). Serious
sequelae in surviving infants include blindness, chronic
lung disease, and cerebral palsy.
Patients with iatrogenic PPROM between 16 and 24
weeks gestation who do not have clinical evidence of intraamniotic infection are candidates for amniopatch therapy.
PPROM is confirmed with a sterile speculum examination
showing vaginal pooling of fluid, ferning, and a positive
Nitrazine test. The maximum vertical pocket of amniotic
fluid is measured sonographically. Patients are placed on
intravenous antibiotics and bed rest for one week to allow
for spontaneous sealing of the membranes. If spontaneous
sealing does not occur, 1 unit of autologous platelets and
cryoprecipitate are prepared if the patient is eligible for
autologous donation. Otherwise, donor platelets and cryoprecipitate are prepared.
After informed consent, an amniocentesis is performed
using a 22 gauge needle. The needle is directed into an
available pocket of fluid regardless of the site of the previous invasive procedure. A K-51 tubing extension
attached to a three-way stopcock is connected to the hub
of the needle. Platelets are administered first, followed by
cryoprecipitate. In our original protocol, 1 whole unit of
platelets was injected. We have subsequently reduced the
dose of platelets to one-half a unit because of an unexplained fetal death demise, an adverse effect probably
caused by sudden activation of a large number of platelets.
In our series of 28 cases, the average gestational age at
the time of the procedure was 19 weeks and 3 days. The
average gestational age of delivery in patients who did not
have an intrauterine fetal demise was 33 weeks and 4 days.
Overall, membrane sealing occurred in 19 of 28 patients
(67.9%). Of the 28 patients treated, 11 had a large membrane detachment but no overt leakage of fluid. The
detachment of the membrane occurred from fluid escaping
TTTS
1. Gestational age of less than 26 weeks.

2. Ultrasound diagnosis of a single monochorionic placenta by ultrasound in the first trimester of pregnancy.
3. Polyhydramnios in the recipients amniotic cavity
with a deepest vertical pool 8 cm or 10 cm before
or after 20 weeks of gestation, respectively.
4. Oligohydramnios in the donors amniotic sac with a
deepest vertical pool 2 cm.
Preoperative evaluation consists of ultrasound examination, including morphological examination, fetal Doppler, cardiothoracic index, identification of placental
location, and cord insertions. Amniocentesis or amniore-
duction prior to laser may cause intra-amniotic bleeding

and therefore make the procedure more difficult, or even
impossible, due to impaired visualization. The site of entry
is chosen as demonstrated in Fig. 1, for the scope to be
entered at a right angle to the long axis of the small twin in
order to maximize the chance to ensure adequate visualization of the placental surface and intertwin membranes.
Ideally, the scope should also be entered alongside a virtual
line joining the two cord insertions. When these criteria are
met, the vascular equator of the placenta as well as the
vascular anastomoses on the chorionic plate are more
likely to be visualized in the operative field.
Prophylactic cefazolin 2 g, indomethacin suppository
100 mg, and oral flunitrazepam are given before surgery
and local anesthesia with nonadrenalinized xylocaine is
injected down to the myometrium. A 10 Fr cannula for a
central venous catheter loaded with a trocar is introduced
percutaneously under continuous ultrasound guidance. A 2
mm 08 fetoscope (Storz 26008 AA) is passed down a straight
or curved sheath to operate on posterior or anterior placentas, respectively. The sheath also has a working channel carrying a 1 mm diode laser fiber.
A systematic examination of the chorionic plate alongside the insertion of the inter-twin membrane is performed. Identification of crossing vessels and of their
arterial or venous nature is possible because arteries cross
over veins and show a darker red color than veins owing to
a lower oxygen saturation in the circulating blood (54).
Selective coagulation of anastomotic vessels is performed
with the aim of separating the monochorionic placenta
into two distinct fetal-placental circulations, sparing the
normal cotyledons of each placental territory. Nonselective coagulation of crossing vessels is only performed
when the distal end or the origin of the vessel cannot
be identified. The power of the diode laser is set at 3050
w. At the end of theprocedure, excessive amniotic fluid is
drained through the sheath of the fetoscope until normal
amniotic fluid volume is obtained with a deepest vertical
pool of 56 cm.
BIPOLAR UMBILICAL CORD OCCLUSION

Selective reduction in complicated monochorionic (MC)
multifetal pregnancies is performed to prevent the delivery
of an anomalous or severely compromised fetus and
improve the perinatal outcome for the surviving co-twin
by delaying delivery or risk associated with spontaneous
loss of the affected. The use of cardiotoxic agents, such as
potassium chloride, is contraindicated in MC pregnancies
because of the potential vascular transmission of the agent
and compromise of the co-twin due to the presence of
placental vascular anastomoses. Thermal vascular occlusive techniques, such as bipolar umbilical cord occlusion
(BPC), have been shown to achieve the stated goals with
minimal maternal morbidity. Indications for BPC that are
unique to MC gestations include twin reverse arterial
perfusion, discordant fetal anomalies, and isolated severe
growth lag. BPC has also been used as a primary intervention in advanced twin-twin transfusion syndrome or as
a secondary procedure when alternative therapies such as
179
amnioreduction or laser have failed to correct the disease

process.
The procedure was originally described by Deprest et al.
(55). In brief, using the standard sterile technique, the
patients abdomen is properly prepared and draped. An
abdominal ultrasound is performed to confirm fetal position, viability, and umbilical cord locations. General and
conduction anesthesia may be used; however, intravenous
sedation with local infiltration of 1% lidocaine or 0.25%
bupivicane for subcutaneous, deep muscle, and fascia
anesthesia is usually sufficient and is associated with less
maternal morbidity. A small skin incision is made to allow
insertion of an endoscopic trocar. Under continuous ultrasound guidance, the instrument is inserted through a
placental free window toward the targeted umbilical cord,
ideally avoiding the gestational sac of the normal co-twin.
Once the trocar is secured in the amniotic sac, the obturator is removed. The bipolar forceps are inserted and
advanced to the umbilical cord.
The cord is grasped and positioned away from the
amnion before thermal energy is applied. The duration
and wattage (W) necessary for occlusion will vary, from
2060 s and 2050 W, respectively, based on the gestational age and umbilical cord thickness. When a full-thicknessgrasp exists, application of the thermal energy will
result in turbulence and streaming of amniotic fluid
adjacent to the forceps. It is not uncommon to have an
audible pop secondary to the heating of Whartons jelly
and subsequent rupture of amnion at the site of occlusion,
which should not be perceived as a sign of completed
coagulation. As a result of the natural spiral of the umbilical cord, complete occlusion of all vessels requires 23
applications of the forceps at adjacent sites. Pulse and color
flow Doppler blood flow studies are performed to confirm
cord occlusion at each site.
The size of the BPC forceps that have been used for these
procedures has varied from 2.25.0 mm. The majority of
procedures have been performed with commercially available single-use 3.0 mm bipolar diathermy forceps (5558).
Intravenous prophylactic antibiotics and indomethacin
for tocolysis are generally given prior to the procedure.
Postoperative monitoring for uterine contractions and,
depending on the gestational age, continuous or intermittent fetal heart rate should be done for at least 2 hours. The
majority of programs will observe patients for an extended
period of 1224 h with limited activity. Subsequent doses of
antibiotics and tocolytic treatment are given during this
time. Prior to discharge, a limited ultrasound is performed
to determine the amniotic fluid volume and assess for signs
of hydrops and anemia, including Doppler velocemitry of
the middle cerebral artery and, where appropriate, similar
studies of the umbilical artery and ductus venosus. If no
evidence of preterm labor, leaking of amniotic fluid, or
bleeding exists, the patient is discharged with instructions
to continue with modified bed rest at home for 710 days,
take her temperature bid, and report an elevation, leaking
of vaginal fluid, bleeding, or contractions. An ultrasound is
performed in 1014 days and then at a minimum every 4
weeks thereafter. Additional ultrasounds and fetal monitoring should be performed as clinically indicated by the
primary disease and gestational age.
180
BIBLIOGRAPHY
1. Harrison MR, Adzick NS. Open Fetal Surgery Techniques.
The Unborn Patient: The Art and Science of Fetal Therapy,
3rd ed. Harrison MR, Evan MI, Adzick NS, Holzgreve W,
editors. New York: WB Saunders Company; 2001. pp 247
255.
2. Harrison MR, Anderson J, Rosen MA, et al. Fetal surgery in
the primate I. Anesthetic, surgical and tocolytic management
to maximize fetal-neonateal survival. J Pediatr Surg
1982;17:115122.
3. Nakayama DK, Harrison MR, Seron-Ferre M, et al. Fetal
surgery in the primate II. Uterine electromyographic response
to operative procedure and pharmacologic agents. J Pediatr
Surg 1984;19:333339.
4. Adzick NS, Harrison MR, Glick PL, et al. Fetal surgery in the
primate III. Maternal outcome after fetal surgery. J Pediatr
Surg 1986;21:477480.
5. Bianchi DW, Crombleholme TM, DAlton ME. Cystic
Adenomatoid Malformation. In: Bianchi DW, Crombleholme
TM, DAlton ME, editors. Fetology-Diagnosis & Management
of the Fetal Patient. New York: McGraw-Hill; 2000. pp 289
297.
6. Bianchi DW, Crombleholme TM, DAlton ME. Sacrococcygeal
teratoma. In: Bianchi DW, Crombleholme TM, DAlton ME,
editors. Fetology-Diagnosis & Management of the Fetal
Patient. New York: McGraw-Hill; 2000. pp 867877.
7. Altman RP, Randolph JG, Lilly JR. Sacrococcygeal teratoma:
American Academy of Pediatrics Surgical Section Survey
1973. J Pediatr Surg 1974;9:389398.
8. Adzick NS. Management of fetal lung lesions. Clin Perinatol
2003;30:481492.
9. Hedrick HL, Flake AW, Crombleholme TM, et al. Sacrococcygeal teratoma: Prenatal assessment, fetal intervention, and
outcome. J Pediatr Surg 2004;39(3):430438.
10. Crombleholme TM, Coleman B, Hedrick H, et al. Cystic adenomatoid malformation volume ratio predicts outcome in prenatally diagnosed cystic adenomatoid malformation of the
lung. J Pediatr Surg 2002;27(3):331338.
11. Rice HE, Estes JM, Hedrick MH, et al. Congenital cystic
adenomatoid malformations: A sheep model. J Pediatr Surg
1994;29:692696.
12. Flake AW. Fetal sacrococcygeal teratoma. Sem Pediatr Surg
1993;2:113120.
13. Adzick NS, Crombleholme TM, Morgan MA, et al. A case
report. A rapidly growing fetal teratoma. Lancet 1997;349:538.
14. Hedrick HL. Ex utero intrapartum therapy. Semi Ped Surg
2003;10(3):190195.
15. Wilson RD, Johnson MP, Flake AW, et al. Reproductive outcomes after pregnancy complicated by maternal-fetal surgery.
Am J Obstet Gynecol 2004;191:14301436.
16. Harrison MR, Filly RA, Parer JT, et al. Management of the
fetus with a urinary tract malformation. JAMA 1981;
246(6):635639.
17. Nakayama D, Harrison M, deLorimier A. Prognosis of posterior urethral valves present at birth. J Ped Surg 1986;21:43
45.
18. Golbus MS, Harrison MR, Filly RA, et al. In utero treatment of
urinary tract obstruction. Am J Obstet Gynecol 1982; 383
388.
19. Berkowitz RL, Glickman MG, Smith GJ, et al. Fetal urinary
tract obstruction: What is the role of surgical intervention in
utero? Am J Obstet Gynecol 1982;144(4):367375.
20. Rodeck C, Nicolaides K. Ultrasound guided invasive procedures in obstetrics. Clin Obstet Gynecol 1983;10:515.
21. Beck AD. The effect of intra-uterine urinary obstruction upon
the development of the fetal kidney. Urol 1971;105:784789.
22. Pringle KC, Bonsib SM. Development of fetal lamb lung and
kidney in obstructive uropathy: A preliminary report. Fetal
Ther 1988;3(1-2):118128.
23. Manning FA, Harman CR, Lange IR, et al. Antepartum
chronic fetal vesicoamniotic shunts for obstructive uropathy:
A report of two cases. Am J Obstet Gynecol 1983;145(7):819
822.
24. Young H, Frontz W, Baldwin J. Congenital obstruction of the
posterior urethra. J Urol 1919;3:289365.
25. Reuss A, Wladimiroff J, Niermeyer M. Antenatal diagnosis of
renal tract anomalites by ultrasound. Pediat Nephrol
1987;1:546552.
26. Johnson MP, Bukowski TP, Reitleman C, et al. In utero
surgical treatment of fetal obstructive uropathy: A new comprehensive approach to identify appropriate candidates for
vesicoamniotic shunt therapy. Am J Obstet Gynecol
1994;170(6):17701776; discussion 17761779.
27. Quintero RA, Hume R, Smith C, et al. Percutaneous fetal
cystoscopy and endoscopic fulguration of posterior urethral
valves [see comments]. Am J Obstet Gynecol 1995;172(1 Pt
1):206209.
28. Quintero RA, Johnson MP, Romero R, et al. In-utero percutaneous cystoscopy in the management of fetal lower obstructive uropathy. Lancet 1995;346(8974):537540.
29. Quintero RA, Shukla AR, Homsy YL, et al. Successful in
utero endoscopic ablation of posterior urethral valves: A
new dimension in fetal urology. Urology (Online) 2000;55(5):
774.
30. Steinbok P, Irvine B, Cochrane DD, Irwin BJ. Long-term
outcome and complications of children born with myelomeningocele. Childs Nerv Syst 1992;8:9296.
31. McLone DG. Continuing concepts in the management of spina
bifida. Pediare Neurosurg 1992;18:254257.
32. Tulipan N, Hernanz-Schulman M, Bruner JP. Reduced
hindbrain herniation after intrauterine myelomeningocele
repair: A report of four cases. Pediatr Neurosurg 1998;29:
274278.
33. Tulipan N, Hernanz-Schulman M, Bruner JP. Intrauterine
myelomeningocele repair reverses preexisting hindbrain herniation. Pediatr Neurosurg 1999;31:137142.
34. Bruner JP, Tulipan N, Paschall RL, Boehm FH, Walsh
WF, Silva SR, Hernanz-Schulman M, Lowe LH, Reed GW.
Fetal surgery for myelomeningocele and the incidence
of shunt-dependent hydrocephalus. JAMA 1999;282:1819
1825.
35. Johnson MP, Sutton LN, Rintoul N, Crombleholme TM, Flake
AW, Howell LJ, Hedrick HL, Wilson RD, Adzick NS. Fetal
myelomeningocele repair: Short-term clinical outcomes. Am J
Obstet Gynecol 2003;189:482487.
36. Bruner JP, Boehm FH, Tulipan N. The Tulipan-Bruner trocar
for uterine entry during fetal surgery. Am J Obstet Gynecol
1999;181:11881191.
37. Mangels KJ, Tulipan N, Bruner JP, Nickolaus D. Use of
bipedicular advancement flaps for intrauterine closure of
myeloschisis: Technical report. Pediatr Neurosurg 2000;
32:5256.
38. Kohl T, Sharland G, Allan LD, Gembruch U, Chaoui R, Lopes
LM, Zielinsky P, Huhta J, Silverman NH. World experience of
percutaneous ultrasound-guided balloon valvuloplasty in
human fetuses with severe aortic valve obstruction. Am J
Cardiol 2000;85:12301233.
39. Tworetzky W, Wilkins-Haug L, Jennings RW, van der Velde
ME, Marshall AC, Marx GR, Colan SD, Benson CB, Lock JE,
Perry SB. Balloon dilation of severe aortic stenosis in the fetus:
Potential for prevention of hypoplastic left heart syndrome:
candidate selection, technique, and results of successful intervention. Circulation 2004;110:21252131.
IONIZING RADIATION, BIOLOGICAL

40. Luton D, de Lagausie P, Guibourdenche J, Peuchmaur M,
Sibony O, Aigrain Y, Oury IF, Blot P. Influence of amnioinfusion in a model of in utero created gastroschisis in the pregnant ewe. Fetal Diagn Ther 2000;15:224228.
41. Luton D. Etude de Linflammation, dans le Laparoschisis,
Humain et dans un Modele de Laparoschisis de Bredis. These;
2001.
42. Luton D, de Lagausie P, Guibourdenche J, Oury IF, Sibony O,
Vuillard E, Boissinot C, Aigrain Y, Beaufils F, Navarro J, Blot
P. Effect of amnioinfusion on the outcome of prenatally diagnosed gastroschisis. Fetal Diagn Ther 1999;14:152.
43. The NICHD National Registry for Amniocentesis Study
Group. Midtrimester amniocentesis for prenatal diagnosis.
Safety and accuracy. JAMA 1976;236:14711476.
44. Rodeck C. Fetoscopy guided by real-time ultrasound for pure
fetal blood samples, fetal skin samples, and examination of the
fetus in utero. Br J Ob Gyn 1980;87:449456.
45. Quintero R, Reich H, Puder K, et al. Brief report: Umbilicalcord ligation of an Acardiac twin by fetoscopy at 19 weeks of
gestation. N Engl J Med 1994;33:469471.
46. Gold R, Goyert G, Schwartz D. Conservative management of
second-trimester post-amniocentesis fluid leakage. Obstet
Gynecol 1989;74:745747.
47. Bengtson J, VanMarter L, Barss V. Pregnancy outcome after
premature rupture of the membranes at or before 26 weeks
gestation. Obstet Gynecol 1989;73:921927.
48. Beydoun S, Yasin S. Premature rupture of the membranes
before 28 weeks: Conservative management. Am J Obstet
Gynecol 1986;155:471479.
49. Rotschild A, Ling E, Puterman M. Neonatal outcome after
prolonged preterm rupture of the membranes. Am J Obstet
Gynecol 1990;162:4652.
50. Sebire NJ, Snijders RJ, Hughes K, et al. The hidden mortality
of monochorionic twin pregnancies. Br J Obstet Gynecol
1997;104(10):12031207.
51. Patten RM, et al. Disparity of amniotic fluid volume and fetal
size: Problem of the stuck twin-US studies. Radiology
1989;172:153157.
52. Danskin FH, Neilson JP. Twin-to-Twin transfusion syndrome:
What are appropriate diagnostic criteria? Am J Obstet Gynecol 1989;161:365369.
53. Senat MV, Deprest J, Boulvain M, Paupe A, Winer N, Ville Y.
Endoscopic laser surgery versus serial amnioreduction for
severe twin-to-twin transfusion syndrome. N Engl J Med
2004;351:136144.
54. Benirschke K, Driscoll S. The Pathology of the Human Placenta. New York: Springer-Verlag; 1967.
55. Deprest JA, Audibert F, Van Schoubroeck D, Hecher K,
Mahieu-Caputo D. Bipolar coagulation of the umbilical cord
in complicated monochorionic twin pregnancy. Am J Obstet
Gynecol 2000;182:340345.
56. Nicolini U, Poblete A, Boschetto C, Bonati F, Roberts A.
Complicated monochorionic twin pregnancies: Experience
with bipolar cord coagulation. Am J Obstet Gynecol
2001;185:703707.
57. Taylor MJO, Shalev E, Tanawattanacharoen S, Jolly M,
Kumar S, Weiner E, Cox PM, Fisk NM. Ultrasound-guided
umbilical cord occlusion using bipolar diathermy for Stage III/
IV twin-twin transfusion syndrome. Prenat Diagn 2002;
22:7076.
58. Johnson MP, Crombleholme TM, Hedrick H, et al. Bipolar
umbilical cord cauterization for selective termination of complicated monochorionic pregnancies. Am J Obstet Gynecol
2001;185(6):S245.
See also FETAL
MONITORING; HYDROCEPHALUS, TOOLS FOR DIAGNOSIS
AND TREATMENT OF; MONITORING, UMBILICAL ARTERY AND VEIN.
181
OIL. See LENSES, INTRAOCULAR.

ION-EXCHANGE CHROMATOGRAPHY. See
CHROMATOGRAPHY.

EFFECTS OF
ZELENNA GOLDBERG
Department of Radiation
Oncology
Davis, California
INTRODUCTION
Radiation biology refers to all biologic responses induced in
cells and tissues by ionizing radiation, a term that encompasses the study of all action of ionizing radiation on living
things. Ionizing radiation (IR) is radiation that has sufficient energy to cause the ejection of an orbital electron from
its path around the nucleus, which indicates that the
photon or charged particle can release large amounts of
energy in a very small space. IR is separated into electromagnetic radiation and particulate radiation. Electromagnetic radiation consists of X rays or g rays, which differ only
in their mode of production, not in their physical effects in
interacting with biologic tissue. X rays are produced by
machines that accelerate electrons and then focus them to
hit a target usually made of tungsten or gold. The kinetic
energy is transferred from the electrons to the target and
then is released as photons. These are the most common
medical exposures through diagnostic or therapeutic radiation. In contrast, g-radiation is produced within the
nucleus from radioactive isotopes. The g rays result from
the unstable nuclear configuration decaying to a more
stable state and releasing the extra energy in this transition in the form of the g ray. Natural background radiation
is of this type. Particulate radiation is those radiation
sources that are not photons and consist of electrons,
protons, a-particles, neutrons, and heavy charged particles
such as the nuclei of carbon, neon, argon, or iron. These
particles are positively charged because the orbiting electrons have been stripped from them.
As noted, IR is defined by its causing the ejection of an
orbital electron when the radiation interacts with tissue.
These ionizing events can be further subdivided by being
directly ionizing or indirectly ionizing. Directly ionizing
events are those in which a charged particle with sufficient
kinetic energy interacts with the tissue, directly resulting
in the ejection of the orbital electron. These events happen
frequently within an exposed tissue, so the charged particle rapidly transfers its energy to the tissue, a concept
codified as linear energy transfer (LET). Charged particles
have a high LET. In contrast, indirectly ionizing radiation
such as g- or X-radiation is first absorbed in the material
and secondary, energetic charged particles (electrons) are
released. Because this latter process only occurs when the
photon is close enough to an atom to interact, it is referred
to as sparsely ionizing radiation and has a low linear
energy transfer (LET).
182
A longstanding paradigm in radiation biology has been

that many effects induced by IR, including its carcinogenic
effects and ability to kill cancer cells, are the result of DNA
damage arising from the actions of IR in cell nuclei, especially interactions of IR and its products with nuclear DNA
(13). When a charged particle or secondary electron produced by the interaction of a photon with an orbital electron damages DNA itself, it is known as the direct action of
IR. Yet, DNA represents a small fraction of the actual size
of the cell. Therefore, it was recognized that most of the
interaction of IR within a cell would be with more abundant
biomolecules, specifically water. If the action is mediated
through the intracellular production of radiolytic reactive
products, e.g., OH, H, O2 , and H2O2, that are generated in
aqueous fluid surrounding DNA, then the DNA damage is
called indirect action. These processes unquestionably can
result in a variety of types of DNA damage, including DNA
single- and double-strand breaks, modifications of deoxyribose rings and bases, intra- and inter-strand DNADNA
cross-links, and DNAprotein cross-links (1,4,5). About a
third of all DNA damage is caused by the direct effects of
sparsely ionizing g and X rays, with the remaining balance
being attributable to the indirect actions of IR. With highLET radiation such as the more densely ionizing a-particles that are emitted by radon and radon progeny, the
direct actions of IR on DNA become more predominant
and the nature of the DNA modifications become much
more complex. Regardless of the type of IR, all of the above
forms of DNA damage can lead to untoward effects in cells
if unrepaired or misrepaired. With specific regard to carcinogenesis, genomic mutations caused by IR are widely
thought to arise from DNA damage that is subsequently
converted into a mutation as a result of misprocessing by
DNA repair mechanisms or that is converted into a heritable mutation when DNA undergoes replication.
This classic view of radiation biology is giving way to a
more complex and complete understanding that the cell
membrane and other intracellular compartments, as well
as the tissue vasculature, are important targets of IR.
Furthermore, as detailed below, although intracellular
effects are better defined, we now recognize that the extracellular environment and cellcell contact play important
roles in modulating the effects of IR at the cellular and
tissue levels.
MOLECULAR RADIATION BIOLOGY/BYSTANDER EFFECTS

IR is a cellular toxin that is sensed at the cellular level
through the ATM-p53-p21 pathway (6,7). Although the
upstream sensor of ATM remains to be definitively elucidated, the initial radiation sensing protein likely has a
redox sensor and undergoes a conformational change, or
possibly is phosphorylated, resulting in the phosphorylation of ATM (8). This, in turn, activates the p53 pathway
leading to cell cycle arrest, nuclear translocation of transcription factors such as NF-kB, and either cellular repair
or apoptosis. How an individual cell commits to a given fate
(repair or apoptosis) remains unclear, but undoubtedly it
represents the final integrated response to many simultaneous intracellular events. Several excellent reviews on
this topic have been written (9,10). It should be noted that

IR also affects the 26s proteosome, which is another level of
non-nuclear intracellular response affecting cell survival,
presumably through alterations in the removal of activated
(phosphorlyated) proteins, or proteins active in apoptotic
processes such as Bcl2 (11,12).
Radiation effects in cells not directly hit by a radiation
ionizing event are called bystander effects. Although initial
interest in this effect can be found in the medical literature
as far back as into the 1950s, the current interest in the
area was stimulated by a study published in 1992 in which
Nagasawa and Little (13) observed increases in the frequency of sister chromatid exchanges in 30% of immortalized Chinese hamster ovary cells that received low-dose
exposure to a-particles, even though less than 1% of the
cells nuclei were estimated to actually receive direct
nuclear hits by an a-particle. That a relatively low percentage of the cells experienced one or more direct hits by the
a-particles, be they in the cytoplasm or in the nuclei,
suggested the possibility that some mechanism was conveying a radiation-associated response to unirradiated
cells. Other groups went on to confirm and extend on this
finding. It is now well recognized that cells do not require a
direct nuclear traversal to result in radiation changes.
There are extracellular responses, predominantly TGF-b
mediated through media (cell culture experiments) or
extracellular fluids (tissues), but other factors cannot be
excluded (14). Furthermore, it is recognized that cellcell
contact and gap junctional communications are critical in
transmitting the signal from the directly irradiated cell to
the neighboring, bystander cells (15).
TIME, DOSE, AND FRACTIONATION

The biologic effects of IR in tissue relate to the size of the
dose delivered, time between radiation exposures and the
total dose of IR given. Although environmental exposures
are chronic and (usually) low level, medical exposures are
acute and can be repetitive when given for the treatment of
cancer. Radiation therapy for the treatment of malignancy
remains the most effective anti-cancer agent discovered,
and treatment schedules are predicated on the 4 Rs of
radiobiology: repair, repopulation, redistribution, and
reoxygenation.
Repair of radiation-induced damage underlies the
intrinsic radiation sensitivity of the cell to radiation cell
killing. Cells can be broadly grouped into those that can
repair significant amounts of damage and those with more
limited repair capacity. The former descriptive category
corresponds to tissue types where there is limited normal
cell turnover, such as lung, kidney, or brain, and these are
tissues that display damage after a more prolonged time
and are therefore known as late responding tissues. The
cells with more limited intrinsic repair capacity are known
as acute responding tissues and are typified by skin or gut.
These cell types have a limited life span and are constantly
being replaced within normal physiologic functioning.
Repopulation refers to the generation of new cells to
replace those killed by the IR exposure. Although therapeutically beneficial for containing normal tissue toxicity,
repopulation is also active in malignant tissue and thereby

allows greater numbers of tumor clonogens to develop after
treatment. For some types of tumors, an acceleration of
repopulation begins after 4 weeks of radiation therapy,
which can compromise clinical outcome if prolonged therapeutic IR fractionation schemes are used.
Redistribution refers to the changes in cell assortment
across the cell cycle. It has long been established that cells
vary in their sensitivity to the cytotoxic effects of IR depending on where in the cell cycle they are at the time of
irradiation (16,17). Cells are most sensitive to IR effects
in the G2/M phase and are most resistant during the Sphase. G1 is intermediate between these two. Thus, clinical
radiation therapy schedules use multiple fractions to overcome the relative resistance of the cells in S-phase of the cell
cycle on a given treatment day. This difference in radioresistance across stages in the cell cycle also underlies one
mechanism of the synergy between radiation therapy and
many chemotherapeutic agents in the treatment of malignant disease. Although S-phase cells resist killing from the
IR, they are more sensitive to some chemotherapeutics that
are active during S-phase when the DNA is most exposed
and is replicating. Furthermore, this alteration of cell cycle
sensitivity to IR cytotoxicity also has been an area of
research for IR-biologic response modifiers. If one can
increase the percentage of cells in G2/M at the time of IR,
then the relative cell kill per fraction of IR will increase.
Reoxygenation refers to the presence or absence of
molecular oxygen within the cell at the time of IR delivery.
As detailed at the beginning, most of the cellular damage
from IR is mediated through the production of free radicals
within the cell. If molecular oxygen is present, these free
radicals can be transformed into more complex peroxide
radicals, which are more difficult for the cell to repair. This
is classically known as fixing the radiation damage in the
British use of the term fix (make more solid), not the
American (repair). The increase in cell killing from IR in
the presence of oxygen versus under hypoxic conditions is
known as the oxygen enhancement ratio, and it is approximately a factor of 23, dependent on where in the cell cycle
the irradiated cell sits. Although normal tissues have a
well-maintained vascular supply so that oxygenation is
essentially constant, tumors have a tortuous and unstable
vasculature, so that the vessels can open and close erratically. This type of hypoxia is referred to as acute
hypoxia. Chronic hypoxia occurs when the tumor outgrows its blood supply and the tumor cells are simply
beyond the diffusion capability of molecular oxygen.
Attempts to exploit this differential, either by sensitizing
the hypoxic cells or by specifically targeting them, have
been explored and remain active areas of research.
Furthermore, identification of the presence of hypoxia
remains a significant clinical strategy (18).
The four Rs of radiobiology reflect intracellular controls
of overall radiation response. The tissue level effects from
IR in the treatment of cancer are dependent on several
parameters: volume, dose per fraction, total dose, and time
between fractions. The volume of tissue irradiated is critical for determining the long-term repair by repopulation
by normal cells to fill in for those irreparably damaged by
the IR. Each cell type has its own intrinsic radiation
183
sensitivity, and the tissue organization (serial or parallel

cellular arrangement) as well as the tissue vasculature
play critical roles in tissue repair and thus radiosensitivity.
Tissues are subdivided into two categories of radiosensitivity based on when they display their damage. Tissues
that display their radiation induced damage during a
standard course of medical radiation therapy are known
as acute responding tissues. These tissues naturally have
a large amount of cellular turnover, such as skin or gut,
and at a cellular level, this corresponds to lesser ability to
repair sublethal DNA damage (i.e., a larger proportion of
DNA damage is lethal and nonrepairable). In contrast,
tissues where there is little or no normal cellular turnover,
such as brain, kidney, lung, or spinal cord, there is a
substantial ability to repair sublethal damage, and tissue
toxicity from radiation therapy is displayed late, long after
therapy has finished. These tissues are therefore labeled
late responding tissues. Therapeutic strategies are
designed to separate these two types of responses as malignant tumors are models of acute responding tissues,
whereas the dose limiting side effects from radiation
therapy are secondary to late responding tissue effects
(19,20).
THE LINEAR NO-THRESHOLD MODEL OF RADIATION
EFFECTS
Based on the data collected from the victims of the bomb
detonations at Hiroshima and Nagasaki, a linear nothreshold model of radiation effects was developed and
adopted for public policy applications. In essence, the
model states that (1) all radiation exposures are biologically active, (2) the response in the cells/tissue is linear
with dose, and (3) there is no threshold below which there is
no or negligible effects. The model was developed from the
moderately low-dose exposure ranges of 0.52 Gy and the
medically significant sequelae of increased mortality (carcinogenic and noncarcinogenic, predominantly cardiovascular) (2123). Hiroshima and Nagasaki data represent the
effects of a single acute dose exposure delivered to the
whole body with the nutritional deprivation from WW2;
as such, they are subject to criticism and questions of their
applicability to modern healthy populations where lowdose radiation exposure is often of a more chronic nature.
Nevertheless, they remain the best available populationbased data, and they have been painstakingly collected and
analyzed. As detailed below, however, the shape of the
response curve at the lowest doses remains controversial.
LOW-DOSE EXPOSURES
Low-dose ionizing radiation (LDIR) in the 110 cGy range
has largely unknown biological activity in the human.
Current modeling for health and safety regulations, as
well as prediction of carcinogenesis, presupposes a linear,
no-threshold model of radiation effects based on the
nuclear bomb explosion data, which estimates the effect
and risk at low dose by extrapolation from measured effects
at high doses. Yet the scientific literature presents a
more complex picture, and few data clearly support a linear
184
dose-response model and none in humans. Numerous studies suggest some effects of LDIR may be benign or even
beneficial under some circumstances (24,25). Reported
benefits include stimulated growth rates in animals,
increased rates of wound healing, reductions in cell apoptosis, enhancements in the repair of damaged DNA, and
increases in radioresistance via the induction of an adaptive response, among others (2628). Other lines of evidence, however, suggest LDIR can be hazardous, and if a
threshold for detrimental responses does exist, it is operational only at very low dose levels, e.g., 1 cGy. Schiestl et
al. (29) found that 1100 cGy doses of X rays could cause
genetic deletions in mice in a linear dose-response manner.
This remains an active area of research (30).
Little is known regarding individual variability in sensitivity to radiation exposure. Studies are actively ongoing
to develop methods to best assess interindividual variability. This understanding will have both a health risk
assessment and medical applications (31).
GENOMIC INSTABILITY
Radiation-induced genomic instability encompasses a
range of measurable endpoints such as chromosome destabilization, sister chromatid exchanges, gene mutation and
amplification, late cell death, and aneuploidy, all of which
may be causative factors in the development of clinical
disease, including carcinoma.
Kadhim et. al. identified the persistence of radiationinduced chromosomal instability following a-particle irradiation in clonal populations of murine bone marrow cells
(32). The same group followed up this seminal work with an
examination of four human bone marrow samples, subjected
to an ex vivo a-particle IR (33). Further research then
demonstrated that the genomic instability phenotype could
be transmitted in vivo when murine hemopoietic cells that
had been irradiated in vitro were transplanted into mice
that had previously had their native bone marrow purged
(34). However, when Whitehouse and Tawn examined
radiation workers in Sellafield, England, who had bone
marrow plutonium deposition evidence for genomic instability was not found (35). Whether g-IR could induce genomic
instability was then examined. The original reports from
Kadhim et al. were negative (32,33), but further research
examining hprt locus mutations convincingly demonstrated
that genomic instability was inducible by g-irradiation (36).
Thus, although genomic instability can clearly be demonstrated in the laboratory, whether it occurs in humans after
IR exposure remains uncertain (37).
Our understanding of the biologic effects of IR is evolving and ever growing. Research has been active in this
field for over 100 years, and it remains a vibrant research
area with the new molecular tools now available. The
target of interest and concern is as small as the individual
DNA base or as large as the whole organism. Mechanistic
studies of the subcellular targets of IR and the cellular
response signaling cascades must be matched with more
complex system evaluations so that the summative effect of
these pathways becomes known. Cell signaling exists
within and between cells, and this cross-talk affects the
ultimate response to IR exposure. Improving our understanding of radiation response at the cellular and tissue
levels will undoubtedly yield advances for medical/therapeutic radiation as well as general cellular biology.
BIBLIOGRAPHY
1. Goodhead DT. Initial events in the cellular effects of ionizing
radiations: Clustered damage in DNA. Int J Radiation Biol
1994;65(1):717.
2. Iliakis G. The role of DNA double strand breaks in ionizing
radiation-induced killing of eukaryotic cells. Bioessays
1991;13(12):641648.
3. Sutherland BM, et al. Clustered DNA damages induced by x
rays in human cells. Radiat Res 2002;157(6):611616.
4. Ward JF. DNA damage produced by ionizing radiation in
mammalian cells: identities, mechanisms of formation, and
reparability. Prog Nucleic Acid Res Mol Biol 1988;35:95125.
5. Sutherland BM, et al. Clustered DNA damages induced in
isolated DNA and in human cells by low doses of ionizing
radiation. Proc Natl Acad Sci USA 2000;97(1):103108.
6. Fernandes N, et al. DNA damage-induced association of ATM
with its target proteins requires a protein interaction domain
in the N terminus of ATM. J Biol Chem 2005;280(15):15158
15164.
7. Kang J, et al. Functional interaction of H2AX, NBS1, and p53
in ATM-dependent DNA damage responses and tumor suppression. Mol Cell Biol 2005;25(2):661670.
8. Lavin MF, et al. ATM signaling and genomic stability in
response to DNA damage. Mutat Res 2005;569(12):123132.
9. McBride WH, et al. A sense of danger from radiation. Radiat
Res 2004;162(1):119.
10. Li L, Zou L. Sensing, signaling, and responding to DNA
damage: Organization of the checkpoint pathways in mammalian cells. J Cell Biochem 2005;94(2):298306.
11. Ghobrial IM, Witzig TE, Adjei AA. Targeting apoptosis pathways
in cancer therapy. CA Cancer J Clin 2005;55(3):178194.
12. Pervan M, et al. Molecular pathways that modify tumor
radiation response. Am J Clin Oncol 2001;24(5):481485.
13. Nagasawa H, Little JB. Induction of sister chromatid
exchanges by extremely low doses of alpha- particles. Cancer
Res 1992;52(22):63946396.
14. Barcellos-Hoff MH. Integrative radiation carcinogenesis:
Interactions between cell and tissue responses to DNA
damage. Semin Cancer Biol 2005;15(2):138148.
15. Goldberg Z, Lehnert BE. Radiation-induced effects in unirradiated cells: A review and implications in cancer. Int J Oncol
2002;21:337349.
16. Brown JM. The effect of acute x-irradiation on the cell proliferation kinetics of induced carcinomas and their normal
counterpart. Radiat Res 1970;43(3):627653.
17. Brown JM, Berry RJ. Effects of X-irradiation on the cell
population kinetics in a model tumour and normal tissue
system: Implications for the treatment of human malignancies. Br J Radiol 1969;42(497):372377.
18. Brown JM, Wilson WR. Exploiting tumour hypoxia in cancer
treatment. Nat Rev Cancer 2004;4(6):437447.
19. Fowler JF. The eighteenth Douglas Lea lecture. 40 years of
radiobiology: Its impact on radiotherapy. Phys Med Biol
1984;29(2):97113.
20. Fowler JF. Potential for increasing the differential response
between tumors and normal tissues: Can proliferation
rate be used? Int J Radiat Oncol Biol Phys 1986;12(4):641 645.
21. Hayashi T, et al. Long-term effects of radiation dose on
inflammatory markers in atomic bomb survivors. Am J Med
2005;118(1):8386.
IONSENSITIVE FIELD-EFFECT TRANSISTORS

22. Preston DL, et al. Effect of recent changes in atomic bomb
survivor dosimetry on cancer mortality risk estimates. Radiat
Res 2004;162(4):377389.
23. Yamada M, et al. Noncancer disease incidence in atomic bomb
survivors, 19581998. Radiat Res 2004;161(6):622632.
24. Loken MK, Feinendegen LE. Radiation hormesis. Its emerging significance in medical practice. Invest Radiol
1993;28(5):446450.
25. Jaworowski Z. Hormesis: The beneficial effects of radiation.
21st Century Sci Tech 1994;7:2227.
26. Shadley JD, Afzal V, Wolff S. Characterization of the adaptive
response to ionizing radiation induced by low doses of X rays to
human lymphocytes. Radiat Res 1987;111(3):511517.
27. Liu SZ, Liu WH, Sun JB. Radiation hormesis: Its expression in
the immune system. Health Phys 1987;52(5):579583.
28. Sagan LA, Cohen JJ. Biological effects of low-dose radiation:
Overview and perspective. Health Phys 1990;59(1): 1113.
29. Schiestl RH, Khogali F, Carls N. Reversion of the mouse pinkeyed unstable mutation induced by low doses of x-rays. Science
1994;266(5190):15731576.
30. Goldberg Z, et al. Effects of low-dose ionizing radiation on gene
expression in human skin biopsies. Int J Radiat Oncol Biol
Phys 2004;58(2):567574.
31. Rocke DM, et al. A Method for Detection of Differential Gene
Expression in the Presence of Inter-Individual Variability in
Response. Bioinformatics. In press.
32. Kadhim MA, et al. Transmission of chromosomal instability
after plutonium alpha-particle irradiation. Nature 1992;
355(6362):738740.
33. Kadhim MA, et al. Alpha-particle-induced chromosomal
instability in human bone marrow cells. Lancet 1994;
344(8928):987988.
34. Watson GE, et al. Chromosomal instability in unirradiated
cells induced in vivo by a bystander effect of ionizing radiation.
Cancer Res 2000;60(20):56085611.
35. Whitehouse CA, Tawn EJ. No evidence for chromosomal
instability in radiation workers with in vivo exposure to
plutonium. Radiat Res 2001;156(5 Pt 1):467475.
36. Kadhim MA, Marsden SJ, Wright EG. Radiation-induced
chromosomal instability in human fibroblasts: Temporal
effects and the influence of radiation quality. Int J Radiat
Biol 1998;73(2):143148.
37. Goldberg Z. Clinical implications of radiation-induced genomic
instability. Oncogene 2003;22(45):70117017.
See also CODES
AND REGULATIONS: RADIATION; NONIONIZING RADIATION,
BIOLOGICAL EFFECTS OF; RADIATION DOSIMETRY FOR ONCOLOGY; RADIATION

PROTECTION INSTRUMENTATION; RADIATION THERAPY, QUALITY ASSURANCE IN.
ION-PAIR CHROMATOGRAPHY. See

CHROMATOGRAPHY.

PAUL A. HAMMOND
DAVID R.S. CUMMING
University of Glasgow
Glasgow, United Kingdom
INTRODUCTION
The pH of a solution is defined as
pH logH
(1)
185
where [H] is the concentration of hydrogen ions in the

solution. The standard laboratory method of measuring the
pH of a solution uses a glass electrode with a thin-walled,
bulb-shaped membrane at the bottom. The electrode is
filled with a standard solution (usually 0.1 M HCl), into
which a silver wire coated with silver chloride is dipped
(Fig. 1). Hydrolysis of both the inside and outside of the
glass membrane forms thin gel layers, separated by dry
glass inside the membrane. Hydrogen ions from the test
solution are able to diffuse into the outside gel layer. The
glass is doped with mobile ions (e.g., Li), which are able to
cross the membrane and relay the concentration of H
ions in the test solution to the inside of the bulb. To make
pH measurements, the potential of the internal silver wire
is measured with respect to a reference electrode. Since the
glass membrane is selective toward [H] ions, the overall
cell potential (at room temperature) is given by the Nernst
equation:
c const
kT
lnH
q
(2a)
const 25:7 103 ln 10 logH
(2b)
const 0:0592 pH
(2c)
where k is the Boltzmann constant, T the absolute

temperature, and q the electronic charge.
A conventional reference electrode consists of a glass
tube containing a filling solution of known composition
(e.g., saturated KCl). In an Ag/AgCl electrode, electrical
contact is made to the filling solution by a silver wire that
has been coated with silver chloride. If saturated KCl is
used, the wire develops a potential of 199 mV versus the
standard hydrogen electrode (SHE). This potential
remains constant as long as the chloride concentration
remains constant. The internal filling solution is separated
from the test solution by a permeable membrane or liquid
junction, usually made from porous glass or ceramic.
Diffusion of ions through the junction provides the conductive path between the reference electrode and the test
solution. The KCl is usually used as the filling solution, as
the mobility of the K and Cl ions are nearly equal,
minimising any build-up of junction potential.
The glass-electrode, with its complicated materials and
internal reference solution, does not lend itself to miniaturization. However, with the arrival of the metal oxide
semiconductor field-effect transistor (MOSFET) in 1960,
another method of measuring the interface potential
became available. The MOSFET is a three-terminal device
in which the voltage on a gate electrode controls the current flowing between source and drain electrodes. The
MOSFET has a metal or polysilicon gate electrode, separated from the bulk silicon by a thin, insulating gate oxide
layer to provide an extremely high input impedance (Fig. 2a).
In an n-channel MOSFET, a positive voltage applied to the
gate electrode attracts electrons from the bulk of the p-type
silicon to the surface beneath the oxide and creates an
inversion region that is rich in mobile electrons. This
inversion region forms a channel that allows current to
flow between source and drain. The minimum gate voltage
186
Glass
electrode
Reference
electrode
AgCl coated Ag wire

Saturated KCl
0.1 M HCl
Glass membrane
Figure 1. Diagram of the glass electrode
together with a silversilver chloride reference
electrode.
that is required to produce strong inversion is termed the

threshold voltage and is given by
VT
FM FS QI QS

2cF
q
CI CI
(3)
The threshold voltage incorporates the work function difference between the metal gate and the silicon (FM FS ),
any fixed charge in the oxide or in the oxidesilicon interface QI and the charge in the depletion region of the silicon
QS . The term 2cF is twice the Fermi level of the p-type
silicon and arises from the definition of strong inversion,
that is the inverted material must have a free-electron
density that is equivalent to the acceptor density in the ptype material (1). The parameter CI is the capacitance of
the oxide (insulator) layer, and all charges and capacitances are expressed per unit area.
In 1970, Bergveld (2) recognized that the MOSFET
structure could be adapted to create an ion-sensitive
FET (ISFET) by omitting the metal gate. He found that
by applying a voltage between the source and drain of this
device and placing it in solution, the current flowing would
vary with the pH of the solution. This first ISFET used the
native gate oxide (SiO2) as the ion-sensing layer and was
sensitive to the concentration of both Na and H ions in
the solution (3). The silicon dioxide, used to insulate the
metal gate from the silicon substrate, has a similar struc-
Gate
Gate
SiO2
SiO2
Source
Channel
Drain
p-substrate
Bulk
(a) MOSFET.
KCl crystals
AgCl coated Ag wire
Liquid junction
ture to the permselective membrane used in the glass

electrode, and will therefore respond to the concentration
of hydrogen ions in a solution. A reference electrode, the
electrical contact of which can be regarded as the gate
terminal, is used to bias the ISFET (Fig. 2b).
A model for an ISFET, made by excluding the metal gate
from a MOSFET, will include all the contributions to its
threshold voltage considered in equation 3. There are an
additional two terms, one due to the potential of the
reference electrode cREF, and the other due to the potential
at the solutionoxide interface Dc xSOL. Here, xSOL is the
constant surface dipole potential of the solvent (usually
water), and Dc is the concentration-dependent interface
potential. The threshold voltage of the ISFET is then
VT cREF Dc xSOL
FS QI QS

2cF
e
CI CI
(4)
since the gate metal of the MOSFET is replaced by the

metal in the reference electrode and its work function has
been included in cREF.
The only term in equation 4 that varies with ionic
concentration is Dc, so measuring the concentration is
simply a matter of measuring VT. The ISFET is then an
ideal transducer with which to measure ionic concentrations since it has an extremely high input impedance, and
hence does not require any bias current to flow in the
solution. It also uses the same fabrication process as a
MOSFET, suggesting that not only can it be made very
small, but that it can be integrated on the same substrate
as the sensor electronics.
THE SITE-BINDING MODEL
Source Current Drain

p-substrate
Bulk
(b) ISFET.
Figure 2. Diagrams showing the similarity in cross-section

between a MOSFET and an ISFET.
Initially, it was assumed that the silicon dioxide insulator

used in the ISFET would obey the Nernst equation, in the
same way as the membrane used in the glass electrode did.
This meant that, for a pH-ISFET, Dc, and hence VT, would
be a linear function of pH with a response of 59 mVpH1.
The model for the ISFET simply substituted the Nernst
equation in place of Dc in equation 4 (4). While this
sD s0 sS 0
cREF cD cREF c0 cD cS c0 cS 0
sS
CI
sD
c0 cD
CH
c0 cS
cD cREF
(6)
(7)
(8)
2kT
sD
sinh1 p
e
8ekTc
Semiconductor
Insulator
IHP
+
+
Solvated
anion
+
+
Solvent
molecule
+
-
Specifically
adsorbed
cation
+
+
Charge density
S
0
Potential
V
EREF
D
0
S
(9)
where k is the Boltzmann constant, e is the permittivity,

and c is the total ionic concentration of the solution. The
last equality (eq. 9) is the GouyChapman model for
the diffuse layer (6), and CH is the capacitance formed
between the inner and outer Helmholtz planes. Combining equations 69 produces:

2kT
sD
s
s
cREF
(10)
D S cS 0
sinh1 p
CH CI
e
8ekTc
|{z} |{z}
cEI Dc
187
(5)
If the potential in the semiconductor bulk is defined to be

zero and the potential of the electrolyte bulk is fixed at
cREF, then
In addition,
OHP
suggested how to operate the ISFET in a circuit, it did not

provide any insight into the chemical processes that occur
at the solutionoxide interface. Nor did it provide an
explanation for the sub-Nernstian and pH dependent sensitivities that were measured for SiO2 ISFETs (5).
The so-called site-binding model assumes that the insulator surface has ionizable sites that react directly with the
electrolyte to bind or release hydrogen ions. The surface
becomes more or less charged depending on the concentration of ions in the solution. The charged surface induces a
layer of complementary charge in the solution that forms
a double-layer capacitance across which the interface
potential is developed. The solution side of the insulator
electrolyte interface is thought to be made up of several
layers as shown in Fig. 3. The solvent molecules and any
ions that are specifically adsorbed onto the surface make up
an inner layer. The locus of the electrical centers of the
adsorbed ions is called the inner Helmholtz plane (IHP) or
Stern layer. The total surface charge density in this plane
due to the ions is s0. Solvated ions in the solution cannot
approach as close to the surface, and their locus of closest
approach forms the outer Helmholtz plane (OHP). The
interaction of the solvated ions with the surface is purely
electrostatic and they are referred to as nonspecifically
adsorbed ions. Because of thermal mixing in the solution,
these ions are distributed throughout the diffuse layer that
extends from the OHP into the bulk of the electrolyte. The
excess charge density in the diffuse layer is sD so that the
total charge in the solution is sD s0.
The charge density in the semiconductor region is sS, so
applying the requirement of charge neutrality:
Electrolyte
cIS
This equation provides a link between the interface

potential Dc and the charge density of the EIS system.
The site-binding model was developed by Yates et al. (7)
to describe the interactions at a general oxideelectrolyte
Figure 3. Model of the electrolyteinsulatorsemiconductor

(EIS) system.
interface. It was later applied to the electrolyteSiO2Si

system by Siu and Cobbold (8) and Bousse et al. (9), whose
approach is outlined here. When an SiO2 surface is in
contact with an aqueous solution, it hydrolyzes to form
surface silanol (SiOH) groups. These groups are amphoteric, meaning that they can react with either an acid or a
base. The acidic and basic character of a neutral SiOH site
is described by the following reactions (Fig. 4) and dissociation constants. (The dissociation constant is the equilibrium constant for a reversible dissociation reaction. It
expresses the amount by which the equilibrium favors the
SiOH
SiOH2
SiOH
SiO
SiOH2
SiO2
Table 1. Values for the Parameters of pH Sensitive

Insulators Found in the Literature
+
+
K1
H
H
SiOH
SiOH
H
H
Insulator
Surface
Solution
Figure 4. Diagram illustrating the surface sites and hydrogenion reactions for SiO2.
products over the reactants.)
SiOH
2 SiOH H ;
SiOH SiO H ;
K1
K2
Reference
18
5 10
8 1018
10 1018
10
1010
104
11
12
12
SiOHH 0
SiOH
2
SiO H 0
SiOH
(11)
(12)
where the square brackets indicate concentrations and the

subscript 0 is used to indicate a surface quantity.
Reactions between surface sites and other ions in the
supporting electrolyte (e.g., Na, Cl) are ignored since
they have been shown to have a negligible effect on the
interface potential (10).
The density of sites on the surface is

N SiOH
2 SiO SiOH
(13)
and the surface charge per unit area is

s0 eSiOH
2 SiO
eDc
ln F
kT
(17)
This equation provides the link between charge, potential,

and pH. The function
q

(18)
F SiOH
2 =SiO
6.2
2:303pHpzc pH
SiOH
NA (sitesm2)
yield
Center
+
10
106
102
K2
SiOH
SiOH2
SiO
1.8
SiO2
Al2O3
Ta2O3
plays a key role in the response of the surface. It can be

written in terms of the normalized net charge
on the
p
surface s
^ 0 s0 =eN, and the parameter d 2 K2 =K1 :
q
s
^ 0 =d ^
s 0 =d2 1 d2 1
(19)
F
1s
^0
Equations 17 and 19 give the solution pH as a function of
both Dc and s0, so now it remains to find the relationship
between Dc and s0. This is done by using the definition of
Dc in equation 10, the charge neutrality condition in
equation 5:
sD s0 Ds sS
(20)
It is usually assumed that Ds 0 (9), so that s0 sD.
Finally, the interface potential Dc can be found as a
function of the solution pH, by using a parametric method
in s
^ 0 . Values of the dissociation constants and surface site
density of SiO2 obtained from the literature are shown in
Table 1, and used to generate the SiO2 pH response curve
in Fig. 5. Not only does the SiO2 surface have a low
sensitivity of 46.3 mV/pH (at pH 7), it also has a
nonlinear response, especially in the acid pH range.
(14)
Due to thermal mixing, the surface concentration of H

ions can be related to the bulk H concentration by Boltzmann statistics:

eDc
(15)
H 0 H exp
kT
as Dc is the potential difference from the electrolyte bulk to
the insulator surface. Multiplying equation 11 by equation
12 and substituting for [H]0 using equation 15 gives

s
p
SiOH
eDc
2
(16)
H K1 K2 exp
kT
SiO

For the
case that Dc 0 and s0 0 i:e:; SiOH
2
p

SiO , we can see from equation 16 that H K1 K2 .
This is the hydrogen ion concentration in the solution
required to produce an electrically neutral surface, and is
called the point of zero charge (pzc). The pH at this point is
denoted pH(pzc) and this can be substituted in 16 to
SiO2
Al2O3
Ta2O5
0.4
Interface potential (V)
188
0.2
5
5.2
0
46
.3
Point of
zero charge
0.2
9.0
0.4
0.6
0
6
8
Solution pH
10
12
14
Figure 5. Graph of the theoretical pH response for SiO2, Al2O3,

and Ta2O5 surfaces.
The insulators Al2O3 and Ta2O5 also have amphoteric

surface groups, so can be modelled using the same parametric method. As shown in Fig. 5, these surfaces produce a
linear response with sensitivities much closer to the Nernstian ideal of 59.2 mVpH1. As a result, Al2O3 and Ta2O5
have been widely used as the pH sensitive layer for fabricating ISFETs. The measured sensitivities for these insulators (5357 mVpH1 for Al2O3, 5657 mVpH1 for
Ta2O5; see Table 2) are close to the theoretical values
shown in Fig. 5.
DEVELOPMENT OF THE ISFET

Much of the subsequent work on ISFETs has concentrated
on measuring pH, as this plays a vital role in many
biochemical systems. Because silicon dioxide has a low
pH sensitivity, the magnitude of which varies with pH
(5), Matsuo and Wise (13) experimented with the use of
silicon nitride (Si3N4) instead. Their ISFET was made by
depositing a layer of nitride on top of the thermally grown
gate oxide. The ISFET was located at the tip of a needleshaped probe, which was covered in a thick layer of SiO2 to
insulate it from the solution (Fig. 6). The ISFET was found
to have an almost ideal pH response and very low sensitivity to sodium and potassium ion concentrations. Selectivity between ion species is important to differentiate
changes in pH from changes in the total ionic concentration
of the solution.
In a sense, silicon nitride was an obvious material to
investigate as it was, and still is, widely used as a passivation layer to protect devices, such as integrated circuits
from the ingress of moisture. Other well-known materials
included the oxides of aluminium and tantalum (Al2O3 and
Ta2O5). The pH sensitivity, ion selectivity, response times,
and drift rates for these materials have been extensively
studied (5). Silicon oxynitride (SiOxNy) and other more
exotic insulators, such as zirconium and tin oxides
(ZrO2, SnO2), and even diamond-like carbon (DLC) have
all been used to make pH ISFETs. Oxides and nitrides of
metals have also been investigated to try and improve
parameters, such as response time and maximum operat-
189
Gate window
Nitride
Oxide
p+
n+
n+
Oxide
Aluminium
p+
p-substrate
Figure 6. Diagram of the cross-section through an ISFET formed
by depositing silicon nitride directly on top of the gate oxide (13).
ing temperature. These include platinum oxide (PtO2),

titanium nitride (TiN), iridium oxide (Ir2O3), and indium
tin oxide (ITO). The published pH sensitivity and drift
rates (where available) for these materials are summarized
in Table 2.
Apart from the choice of pH sensitive material, development of the ISFET has mostly focused on achieving
compatibility with the CMOS process. CMOS devices use
complementary pairs of n-type and p-type MOSFETs to
implement circuits. The CMOS process is the dominant
technology for integrated circuits (ICs), so achieving compatibility would allow complex devices containing ISFETs
to be fabricated by a standard, industrial process. Figure 7
is a simplified cross-section of a CMOS invertor, showing
the polysilicon gate electrodes, the source and drain
regions and the metal interconnections.
ISFETs have been made using both NMOS and PMOS
transistors. However, in a p-substrate process, the bulk
terminal of a PMOS ISFET can be biased above the substrate ground potential, permitting more flexibility in circuit design. The reverse is true for an n-substrate CMOS
process. It has also been shown that the noise performance
of an ISFET is dominated by 1/f or flicker noise (23),
which is lower in PMOS transistors (24).
Initial attempts to integrate ISFETs involved significant modifications to the standard CMOS process. Wong
and White (15) followed the standard sequence of CMOS
process steps, until the metallization stage. They then
removed the oxide, the polysilicon gate electrode, and
the gate oxide above the ISFETs by wet etching. A thinner
Table 2. Values of pH Sensitivity and Drift Rates for ISFETs Made Using a Arange of Materials, Obtained
from the Literature
Material
pH Range
Sensitivity, (mVpH1)
Drift, (mVh1)
Reference
SiO2
410
Unstable
SiOxNy
28.3
49
110
113
210
113
113
112
210
212
1.6810.01
310
2535 (pH < 7)

3748 (pH > 7)
57.4 0.4
1820
40.5 4.0
4556
50
5357
5657
5459
5558
58
59
59.5
0.8 (pH 7)
Not mentioned
0.5 (pH 6.86)
1.0 (pH 7)
Slow response
0.10.2 (pH 7)
0.10.2 (pH 7)
3 mV/h (pH 3)
Not mentioned
Not mentioned
<1
Unclear
14
15
16
5
17
5
5
18
19
20
21
22
PtO2
Si3N4
ZrO2
Al2O3
Ta2O5
DLC
SnO2
ITO
TiN
Ir2O3
190
NMOS
PMOS
Aluminum
Oxide
Polysilicon
S
n+
D
n+
D
p+
S
p+
n-well
p-substrate
Figure 7. Diagram of the cross-section through a CMOS invertor.
oxide layer was regrown and the sensing layer of Si3N4 or

Ta2O5 was deposited on top of this. The contact windows
were then opened and the aluminum deposited to form the
interconnects as for a normal CMOS process.
Bousse et al. (25) took a similar approach, but completed
the CMOS process before etching away the deposited SiO2
above the polysilicon gate of the ISFET. They then deposited Si3N4 over the whole wafer so that the polysilicon was
retained as a floating electrode in the ISFET. This floating
electrode did not reduce the ISFET sensitivity to pH, and
had the additional benefit of making the ISFET less sensitive to changes in light levels. It does this by shielding the
channel from photons that can generate electronhole
pairs, which contribute to the ISFET current. However,
since the nitride they used was deposited by low pressure
chemical vapor deposition (LPCVD) at 785 8C, the aluminium interconnect had to be replaced with tungsten silicide, which was able to withstand this high temperature
step. This meant that a specially modified CMOS process
had to be used to fabricate the ISFETs.
Bausells et al. (26) extended the floating electrode idea
to a two-metal CMOS process by connecting the polysilicon
gate and both metal layers together. In addition, they used
the silicon oxynitride passivation layer as the pH sensitive
material for the ISFET (Fig. 8). This meant that the ISFET
could be fabricated by a commercial foundry using a standard CMOS process, without the need for any process
modifications. The fabricated ISFETs had a sensitivity of
47 mVpH1 and a lifetime of > 2 months. As well as the
advantages of using of a well-characterized industrial
process, there are additional system-on-chip design ben-
H+
H+
H+
H+
H+
H+
H+
H+
H+
H+
H+
H+
Sensing area
H+
Oxynitride passivation layer
Floating electrode
Aluminum
G
Oxide
S
n+
G
D
n+
G
S
n+
D
n+
p substrate
Figure 8. Diagram of the cross-section through an ISFET

fabricated by an unmodified commercial CMOS process (26).
The floating electrode allows the passivation layer to be used as
a pH sensitive insulator.
efits. These include access to libraries of components such

as amplifiers and digital gates that greatly ease the design
of the whole system.
Unfortunately, the unmodified CMOS ISFETs were
found to have large threshold voltages; they were also
based on silicon oxynitride, a material that has been found
to have a widely varying sensitivity, depending on the
deposition conditions (Table 2). The large threshold voltage
has been shown to be caused by trapped charge left on the
floating electrode during fabrication (27). To avoid these
problems, Jakobson et al. (28) removed the passivation
layer by using the aluminum of the floating electrode as
an etch-stop. They then experimented with low temperature evaporation of pH sensitive layers onto the exposed
aluminium. The best performance was obtained by using a
layer of platinum (to prevent the aluminum from corroding), followed by a layer of tantalum oxide.
The floating-electrode ISFETs can be considered as
special cases of the extended-gate ISFET that was first
proposed in 1983. The idea was to separate the electronics
from the chemically active region, and by doing so make the
device easier to passivate and package than a standard
ISFET with an exposed gate insulator. A coaxial polysilicon
structure was used to provide a screened connection
between the chemically sensitive area and the gate of a
MOSFET (29). A more recent CMOS extended-gate ISFET
design used a long (unscreened) aluminium track with one
end connected to the polysilicon gate and the other exposed
by the final pad-etch step (21). This idea has been taken to its
limit by using a discrete, off-the-shelf MOSFET and connecting the gate terminal to the pH-sensitive material with
a length of wire (20). This method is clearly not applicable to
a sensor system-on-chip design, but it does provide a simple
method of characterizing the behavior of the material.
The floating-electrode ISFET has also been used to
protect against electrostatic discharge (ESD). In the first
ESD-protected devices, the polysilicon gate was left intact
and connected to a terminal via a MOSFET switch. This
provided a reverse-biased diode between the floating electrode and the substrate that would breakdown (reversibly)
before the gate insulator was damaged (30). However,
current leakage through the off MOSFET was such that
any response to changing pH decayed to zero in a matter of
seconds. To achieve a steady-state response, a large platinum electrode was connected to the ISFET gate to supply
current from the solution to replace that being lost though
the MOSFET. To avoid the problem of leakage current
altogether, ESD-protected ISFETs have been fabricated
with a separate platinum ring electrode around the sensitive gate area (31). The platinum electrode is a preferential
discharge path to the substrate, protecting the ISFET in the
same manner that a lightning conductor protects a building.
ISFETs have also been adapted to create chemically
modified FETs (CHEMFETs), which are sensitive to the
concentration of ions other than hydrogen. This is achieved
by attaching a polymer membrane containing a suitable
ionophore to the pH sensing surface of the ISFET. The
stability of the ISFETmembrane interface is improved by
the addition of an intermediate hydrogel layer. In this way,
CHEMFETs sensitive to K (32), Na (33), and other
cations have been developed.
Bond-wire
Metal contact
Track
D
PCB
p-substrate
Epoxy
PACKAGING
One of the main obstacles that has prevented the commercialization of ISFET based devices is the repeatability and
reliability of the encapsulation procedure. It is normal for
the encapsulant to be applied by hand, covering the chip
and bond wires, but leaving a small opening above the
sensing area. Epoxy is the most extensively used material
although it is important to select a composition that is
stable, a good electrical insulator, and does not flow during
encapsulation. Many commercially available epoxies have
been assessed for their suitability by making measurements of their electrical impedence over time (3437).
By using ultraviolet (UV) curable polymers, it is possible
to increase the automation of the packaging process using a
standard mask aligner. A lift-off technique was developed
using a sacrificial layer of photosensitive polyimide to
protect the ISFET gates. Alumina-filled epoxy was applied
by screen printing and partially cured, before the polyimide
was etched away leaving a well in the epoxy (38). After
10 days in solution, leakage currents of 200 nA were
observed. Better results were achieved by direct photopolymerization of an epoxy-based encapsulant. The ISFETs
packaged using this method showed leakage currents of
35 nA after 3 months in solution (38). To avoid polarizing
the reference electrode, a leakage current of < 1 nA is
desirable (5). This photolithographic patterning of the
encapsulant was done at the wafer-level, to all the devices
simultaneously. Subsequently, the wafer was diced up and
the individual chips were wire-bonded and coated with
more encapsulant by hand. At the chip-level, wire-bonded
ISFET chips have been covered (again by hand) with a
0.51 mm thick photosensitive, epoxy-based film, then
exposed and developed (39).
Photoresist 2
Photoresist 1
Polyimide
191
Figure 9. Diagram of the cross-section through

an ISFET-based device in a recessed PCB,
encapsulated using a layer of polyimide and
two layers of photoresist (34).
Some degree of automation was introduced by Sibbald

et al. (34) who used a dip-coating method to apply the
polymers. They first recessed the chip into a PCB and wirebonded the connections, before coating it with a layer of
polyimide. Two layers of photoresist followed, before the
underlying polyimide was etched away (Fig. 9). The packaged devices showed < 10 pA leakage current after 10 days
in solution. However, the encapsulation did exhibit
electrical breakdown for applied bias voltages in excess
of 1.52 V, which was attributed to the high electric field in
the thin layer of resist covering the bond wires. In a
separate study, photosensitive polyimide has also been
used to create the wells that separate the ion-selective
membranes on a multiple ISFET chip (40).
The structure of the ISFET has also been modified to
improve the lifetime and ease of manufacture of the packaged device. One solution was to make the ISFET chip
long and thin (1.2 12 mm) with the sensing area at one
end and the bond pads at the other so that the bond-wires
did not enter the solution (14). The chip itself was encapsulated with a thick layer of silica. More radical solutions
involved bulk micromachining to form back-side contacts
to the ISFET so that the bond wires were on the opposite
side of the chip to the solution. The front side of the chip is
protected by anodic bonding of glass (Fig. 10). A review of
back-side contact ISFETs is provided by Cane et al. (41),
but the technique is not particularly suited to a CMOS
chips, which have many bond-pads arranged around the
perimeter.
ISFET CIRCUITS
When an ISFET is placed in solution, a concentrationdependent potential (Df) is formed at the interface
Gate (RE)
Glass
Liquid
pH-sensitive material
Glass
Oxide
n+
n+
Silicon
p
Metal
Source
Bulk
Drain
Figure 10. Diagram of the cross-section through a

back-side contacted ISFET chip with liquid and
electrical contacts on opposite sides (41).
VS = 0 V
D
G
B
S
Drain current (mA)
3.
192
2.
=
VD
.5 V
VD = 1
VD = 0.5 V
0
1.0
0
1.0
Gate (RE) voltage (V)
(a) Symbol.
(b) IV characteristics.
Figure 11. Diagram of the circuit symbol and a graph of the IV

transfer characteristics (4) for an n-type ISFET.
between the gate insulator and the solution. This potential

modulates the threshold voltage (VT) of the ISFET (eq. 4),
an internal quantity that cannot be directly measured.
Some circuitry is therefore required to convert changes
in the threshold voltage into changes of a measurable
signal. The solution to this problem lies in the use of
feedback control to maintain a constant drain current in
the ISFET. For a FET, operating in the linear region3, the
drain current is given by:
"
#
2
VDS
0W
VGS VT VDS
(21)
ID k
L
2
where k0 is a process-dependent constant, and W, L are
the width, length of the device. (In the linear region,
0 < VDS (VGS VT) and ID varies with VDS.) The parameters VGS and VDS are, respectively, the gate-source and
drain-source voltages applied to the FET. For an ISFET, a
reference electrode in the solution acts as the gate terminal
as shown symbolically in Fig. 11a. The ID vs. VGS curves for
an ISFET as measured by Moss et al. (4) are shown in
Fig. 11b. It is clear from this graph that biasing an ISFET
at a constant drain current is only possible if both VDS and
VGS ( VG VS) are maintained constant. If a reference
electrode is used to control VG, then from equation 21, as VT

changes with ID held constant, VS must change by an equal
and opposite amount to compensate. Measuring Dc (and
hence pH) then becomes a straightforward matter of measuring the terminal voltage VS.
In his original paper on the operation of the ISFET,
Bergveld (3) stated that one of the important advantages of
ISFETs, compared with conventional pH electrodes, is that
there is no need for a reference electrode. Instead, he used a
feedback circuit to control the bulk terminal of the ISFET
and maintain a constant drain current. However, this
makes the assumption that the solution is perfectly isolated from the ISFET source and drain terminals, as well
as the circuit. Any current (even leakage current through
the packaging), that flows into the solution will affect its
potential. To a bulk-feedback circuit, this will be indistinguishable from a change in solution concentration. It is
therefore safer to assume that the solution is grounded, or
at least at some well-defined potential with respect to the
circuit, and use a reference electrode to ensure that this is
the case. For this reason, all of the subsequently published
circuits relating to ISFETs include a reference electrode.
The probe fabricated by Matsuo and Wise (13) contained
an ISFET and a MOSFET of identical dimensions. The
ISFET was configured as a source follower with the MOSFET acting as a constant current source. A saturated
calomel electrode (SCE, shown in Fig. 1) was used as a
reference, and the output voltage measured at the source
terminal of the ISFET. Bergveld (42) used a grounded
reference electrode to avoid the problem of a short circuit
if the solution is already at ground potential (e.g., in an
earthed metal container). He used an instrumentation
amplifier arrangement to maintain a constant current at
a constant drain-source voltage (Fig. 12). Amplifiers A1 and
A2 set VDS as determined by the fixed current flowing in R1.
Current feedback from amplifier A4 adjusts the drain
voltage via R2 to keep the current constant as the threshold
voltage changes. One disadvantage of this circuit is that
the output (measured across R9) is not referenced to a fixed
voltage such as ground.
There have been many other circuit topologies proposed
to keep the drain current and/or the drain-source voltage
constant. A straightforward circuit that achieves both of
R5
A1
R6
R3
VREF
RE
R1
A3
ISFET
R4
A2
Figure 12. Circuit diagram of an ISFET source and

drain follower circuit used to maintain a constant
drain current at a constant drain-source voltage (42).
R2
A4
R9
R7
R8
VOUT
Source
VOUT
Sink
Figure 13. Circuit diagram of constant drain current, constant

drainsource voltage ISFET bias circuit (43).
these objectives was presented by Ravezzi and Conci (43).

It used a pair of unity-gain, noninverting operational
amplifiers (op-amps) to ensure that the voltage dropped
across a resistor is also dropped across the ISFET (Fig. 13).
Constant current was maintained in the resistor by a
current source, and in the ISFET by a current sink. This
arrangement allows the source and drain potentials to
adjust as the threshold voltage changes, allowing the
reference electrode to remain at a fixed potential.
The first integrated ISFET circuit was the so-called
operational transducer published by Sibbald (44) in
1985. It used an ISFET and a MOSFET with identical
geometries as the active devices in the long-tailed pair
input stage of an amplifier. Feedback was used to control
the gate voltage of the MOSFET to ensure that the same
current flowed in both devices. The MOSFET gate voltage
tracked that of the ISFET and so measured the changes
in threshold voltage. The key advantage of this circuit was
that changes in temperature affected both devices equally
and were canceled out.
On-chip
Wong and White (15) recognized that there was little to

be gained from an integrated, miniaturized sensor if it
relied on a large, external reference electrode. Instead,
they used an on-chip gold contact as a quasi-reference
electrode (qRE) and a differential circuit to make measurements between two ISFETs with different pH sensitivities.
The potential difference between the solution and the qRE
will depend on the solution composition. However, like
temperature, this is a common-mode signal, which will
affect both ISFETs equally. Hence, it can be rejected by
means of a differential circuit. Tantalum oxide and silicon
oxynitride were used as sensing layers for the two ISFETs,
which formed the input stages of op-amps integrated onto a
CMOS chip (Fig. 14a). The outputs from the two op-amps
were fed into an off-chip differential amplifier. The overall
circuit gave a response of 4043 mVpH1. The benefit of
the differential approach can be seen in Fig. 14b, which
shows the single-ended (VO1 and VO2) and differential
(VOUT) responses as electrical noise was deliberately
applied to the solution. Two copies of the bias circuit in
Fig. 13 have also been used to create a differential system
with Si3N4 and SiO2 ISFETs (45).
The concept of integration has been extended by Hammond et al. (46) who created a complete digital pH meter on
a single CMOS chip. This design makes use of the libraries
of components provided by the CMOS foundry to integrate
not only the ISFET, but also analog bias circuits, digital
signal processing, and storage onto the same chip (Fig.
15a). The chip was mounted in a recessed PCB and covered
with a thick layer of photoresist so that only the ISFET
area was exposed. The digital response of the device to the
changing pH of the solution in which it is immersed is
shown in Fig. 15b.
MINIATURE REFERENCE ELECTRODES

The first attempt to incorporate a reference electrode on an
ISFET chip used a thin-film Ag/AgCl electrode (47). Electrodes like this, with no internal reference solution, are
sensitive to changes in concentration of their primary ion
(in this case Cl), and are referred to as quasi-reference
electrodes. To solve this problem, Smith and Scott (48) also
Off-chip
SiOxNy ISFET
R2
R1
VOUT
R1
Ta2O5 ISFET
VO2
Output voltages (mV)
VO1
VO1
10
VO2
20
30
VOUT
R2
0
(a) Circuit diagram.
193
15
30
Time (s)
45
60
(b) Output voltage.
Figure 14. Circuit diagram and graph of output voltage for a differential integrated amplifier
based on Ta2O5 and SiOxNy ISFETs (15).
194
Photoresist well
1 mm
ISFET
Memory
Analog
circuits
Microchip
PCB track
PCB
Signal processing
Bond-wire
Recess
(a) Chip layout

1
2
$BF
3
4
$7F
5
Solution pH
pH meter output value (hex)
$FF
$3F
7
$00
0
6
Time (min)
10
12
integrated the reference solution and porous membrane

into the chip. Wells were etched into the back-side of a
silicon wafer, to leave a membrane 1070 mm thick. The
membranes were anodized to porous silicon in a bath of
concentrated hydrofluoric acid. The wells were filled with
saturated KCl and sealed with glass coverslips that had
been coated with thin films of Ag/AgCl. The reference
electrode exhibited a low drift rate of 80 mVh1 (worstcase) and a lifetime of > 2 weeks. However, a method of
mass producing an integrated, liquid-filled electrode has
yet to be developed.
Recent developments of miniature reference electrodes
have focused on the use of polymer-supported electrolyte
gels, to replace the liquid filling solution. Suzuki et al. (49)
developed an electrode that uses finely ground KCl powder
supported in a matrix of poly(vinylpyrrolidone) (PVP). An
exploded diagram of the electrode is shown in Fig. 16. First,
a layer of silver was evaporated onto a glass substrate,
inside a U-shaped gold backbone. A layer of polyimide was
applied to protect the silver and to define the electrode
structure. The AgCl was grown through a slit in the polyimide, and a liquid junction was formed by casting a
hydrophilic polymer into the square recess. The electrolyte
layer, containing the KCl powder, was then screen-printed
over the AgCl and the liquid junction. Finally, a passivating layer of silicone rubber was applied. The electrode can
be stored dry, and activated when required by the injection
of a saturated solution of KCl and AgCl through the
silicone. This miniature reference electrode showed a stability of 1.0 mV over a period of 100 h. No difference was
observed between experimental data obtained with the
miniature reference electrode and with a large, commercial
reference electrode.
(b) pH response
Figure 15. Chip layout and pH response of a single-chip digital
pH meter.
REFERENCE FETs
The sensitivity of the differential circuits already discussed
can be increased if one of the devices has no response to
Electrolyte layer
Silicone layer
Polyimide layer
Silver layer
Liquid junction
slit
Pad
End of junction
Au/Cr layer
Figure 16. Exploded diagram showing the
construction of a miniature Ag/AgCl reference
electrode based on a polymer-supported electrolyte
gel (49).
AgCl
Glass substrate
ISFET
REFET
Epoxy
Oxide
Silicon
300
Glass capillary
Substrate
Figure 17. Diagram of the cross-section through a REFET
created using encapsulated pH buffer solution (50).
changes in pH. Such a device is called a reference FET

(REFET). The first REFET was presented by Comte and
Janata (50) in 1978. It consisted of an ISFET surrounded
by a well of epoxy that was filled with a buffered agarose
gel. A glass capillary was inserted into the gel and sealed in
place with more epoxy (Fig. 17). This acted as a liquid
junction between the internal reference gel and the external solution. The response of the REFET was only
3 mVpH1 and it provided temperature compensation of
0.01 pHC1 when used in a differential arrangement.
However, the techniques required to prepare this REFET
are not well suited to mass production.
The pH sensitivity of an ISFET is due to the presence of
chemical sites on the surface of the insulator (Fig. 4) that
can exchange hydrogen ions with the solution. Attempts to
reduce the density of these sites, and hence the sensitivity,
by chemical modification of the surface proved unsuccessful (51). Instead a thin membrane of parylene was used to
cover an ISFET and convert it into a REFET (52). Parylene
has an extremely low density of surface sites and the
REFET was found to have a very low pH sensitivity of
only 0.5 mVpH1. However, parylene forms an insulating
(or ion-blocked) layer that affects the electrical properties
of the underlying ISFET. Even a very thin membrane
reduces the gate capacitance, and hence transconductance,
dramatically. This is a problem for differential circuits,
which rely on ISFET and REFET having well-matched
electrical properties. If a membrane could be found whose
ion-conducting properties were sufficient to pass the electrical voltage of the solution to the sensing layer of the
underlying ISFET, the transconductance would not be
changed. Clearly, such ion-unblocking membranes must
be insensitive to variations in pH.
Bergveld et al. (53) investigated several candidate polymers for REFET membranes and found that polyacrylate
gave the best performance. An intermediate layer of buffered poly(hydroxyethyl methacrylate) (p-HEMA) hydrogel
was necessary to fix the interface potential and to improve
the adhesion of the membrane (54). The REFET showed
< 2 mVpH1 sensitivity, good mechanical properties, and
its transconductance matched that of the ISFET. Despite
these useful properties, the acrylate membrane was selectively permeable for cations (e.g., potassium ions). This
problem was solved by optimizing the amount of didodecyldimethylammonium bromide (DDMAB) added to the
membrane (Fig. 18a). This large immobile cation repels
mobile cations (e.g., potassium), from the membrane.
Figure 18b shows the performance of the ISFETREFET
Change in threshold voltage (mV)
Buffered agarose
195
250
9 105 mol S1
200
2 105 mol g1
150
100
0 mol g1
50
0
105
104
103
102
101
K+ concentration (mol)
Figure 18. Graphs of the response of an acrylate REFET, and an
ISFET, to changes in potassium and hydrogen ion concentrations
(53).
differential system, as compared to the individual devices,

measured using a platinum qRE.
An alternative approach was developed by van den
Vlekkert et al. (55) who used a thick layer of p-HE
MA hydrogel to produce an ISFET with a retarded pH
response. By controlling the diffusion coefficient of hydrogen ions, a pseudo-REFET with a nonresponse time of 10
min was created. Such a REFET is only really useful in a
flow-through system where the analyte is pumped through
in short bursts followed by rinsing and calibration solutions (34). The p-HEMA hydrogel was also used by Chudy
et al. (56) as the base layer for chemically sensitive FETs
(CHEMFETs) and REFETs. The CHEMFET uses an additional ion-selective membrane containing an ionophore,
selected to target the ion of interest. Exclusion of the
ionophore from the membrane enabled the creation of a
REFET that showed almost no response to pH, potassium,
or calcium ions.
APPLICATIONS
According to Bergveld (57), 20 companies have commercialised ISFETs based on 150 patents. Product offerings
and specifications vary, but in general terms ISFETs have
found applications where small size, fast response, robustness, wide operating temperature range, and operation in
nonaqueous environments are desirable. The ISFET can
also be stored dry, making it more convenient to use than
traditional glass electrodes, since the lifetime of a stored
ISFET is almost indefinite. ISFETs are used for high
resolution applications and products are available with
specified resolutions < 0.01 pH units. However, many of
the devices on the market have a resolution of 0.1 pH
units.
The ISFET has made a significant impact on a number
of industries where it is a requirement to monitor and
control the addition of reagents. It made an early appearance in the food industry where its robust construction
allowed measurement to be made in foodstuffs. In addition,
196
its ability to operate in nonaqueous environments is an

advantage when working with meat and dairy products.
The ISFET has also found applications in the pharmaceutical industry, where a fast re-sponse and high operating temperatures are required. It has also been used in
electroplating where it is able to withstand the corrosive
plating solutions. In addition to manufacturing industries,
the ISFET is also used in waste water management and in
the treatment of effluent.
Early researchers considered medical applications ranging from monitoring the pH in the mouth (particularly in
the context of tooth decay), to direct measurement of ionic
balance in the blood. ISFETs been built into dental prosthetics to enable the direct measurement of pH in the
presence of plaque (58). Recent data on tooth decay was
obtained using a pH imaging microscope (59) although,
unlike the ISFET device, this does not allow for in situ
observations to be made. ISFETs are also used indirectly
for dental applications, for example, in the evaluation of
potential prosthetic materials (60). The ISFET has been
modified in a manner similar to that for the REFET for use
in blood analysis beyond measuring ionic concentrations.
In this embodiment, the ISFET can be made into a sensitive enzyme sensor. In their paper, Sant et al. (61) demonstrated a sensor for creatinine based on an ISFET with a
sensitivity of 30 mVpCreatinine1. The application is in
renal failure and haemodialysis in particular.
Recently, there has been a growth of interest in the use
of ISFETs in a device known as a diagnostic pill. The
concept of a diagnostic pill was developed in the 1950s
and 1960s (62). Such devices are small enough to be
swallowed by a patient, and once inside the gastrointestinal tract, can measure and wirelessly transmit data to an
external receiver over a period of many hours or even days.
The earliest devices were used to measure pressure
changes in the gut, but glass-electrode-based pills were
also made to measure gut pH. The diagnostic pill has made
something of a comeback in recent years, particularly with
the invention of the video pill (63), a device that is capable
of wirelessly transmitting images of the gut with considerably less patient discomfort than would be caused by an
endoscopic procedure. Various new examples of pH measuring pills have also been developed. The Bravo capsule,
which does not actually use an ISFET, is designed for use in
the esophagus (64). During operation it is pinned to the
lining of the esophagus so that it can monitor conditions
such as gastro-esophageal reflux disease (GERD) over a
period of 23 days. The IDEAS capsule (65) that does use
an ISFET, has been built for use in the lower gastrointestinal tract. The device is designed pass naturally through
the gut with as little intervention as possible and is
required to be able to operate in difficult conditions where
biofouling of the sensor is a potential problem.
CONCLUSIONS
The ISFET first appeared in 1970 as an offshoot of the
rapidly growing capability of the microelectronics industry. In its simplest form, the ISFET is a modification of the
traditional MOSFET in which the gate electrode is
removed and the gate oxide exposed to solution. The

ISFET uses changes in the surface chemistry of the gate
oxide to modify the threshold voltage of the transistor and
produce an electronic signal. The sensitivity and performance of the ISFET is highly dependent on the material
used to form the gate oxide layer. The behavior of such
materials is best modeled using a site-binding approach to
calculate the variation in surface charge with solution pH.
A wide variety of materials have been experimented with,
those found to give the best performance are the oxides of
aluminium and tantalum. However, in a standard CMOS
manufacturing process, the passivation layer is made of
silicon nitride, which is also a good pH sensitive material.
It is therefore possible to make good ISFETs using a
standard foundry process with little or no additional
effort. As a result it has become possible to implement
integrated ISFET circuits. A number of circuit topologies
for detecting the change in ISFET threshold voltage have
been designed, those using voltage followers to maintain
the ISFET bias conditions produce the best results.
Further development of ISFET integrated circuits has
enabled complete instruments to be fabricated on a single
IC. To avoid the use of a bulky reference electrode, differential circuits using matched pairs of ISFET and REFET
have been designed. However, difficulties in creating a
good REFET have increased interest in developing miniature reference electrodes that are compatible with IC
processing. The ISFETs have already found widespread
application in manufacturing industries, environmental
monitoring and medicine. It is expected that with
improved miniaturization, integration, and packaging
technologies, new applications will emerge.
BIBLIOGRAPHY
1. Streetman BG, Banerjee S. Solid State Electronic Devices.
5th ed. New Jersey: Prentice Hall; 2000.
2. Bergveld P. Development of an ion-sensitive solid-state device
for neurophysiological measurements. IEEE Trans Biomed
Eng 1970;BM17:70.
3. Bergveld P. Development, operation, and application of ionsensitive field effect transistor as a tool for electrophysiology.
IEEE Trans Biomed Eng 1972;BM19:342.
4. Moss SD, Johnson CC, Janata J. Hydrogen, calcium, and
potassium ion-sensitive FET transducerspreliminaryreport. IEEE Trans Biomed Eng 1978;25:4954.
5. Matsuo T, Esashi M. Methods of ISFET fabrication. Sens
Actuator 1981;1:7796.
6. Bard AJ, Faulkner LR. Electrochemical MethodsFundamentals and Applications. Hoboken (NJ), John Wiley; 2001.
7. Yates DE, Levine S, Healy TW. Site-binding model of electrical
double-layer at oxide-water interface. J Chem Soc Faraday
Trans 1974;70:18071818.
8. Siu WM, Cobbold RSC. Basic properties of the electrolyteSiO2-Si system-physical and theoretical aspects. IEEE Trans
Electron Devices 1979;26:18051815.
9. Bousse L, de Rooij NF, Bergveld P. Operation of chemically
sensitive field-effect sensors as a function of the insulatorelectrolyte interface. IEEE Trans Electron Devices 1983;30:
12631270.
10. Bousse L, de Rooij NF, Bergveld P. The influence of counterion
adsorption on the c0/pH characteristics of insulator surfaces.
Surf Sci 1983;135:479496.

11. Harame DL, Bousse LJ, Shott JD, Meindl JD. Ion-sensing
devices with silicon-nitride and borosilicate glass insulators.
IEEE Trans Electron Devices 1987;34:17001707.
12. van Hal REG, Eijkel JCT, Bergveld P. A general model to
describe the electrostatic potential at electrolyte oxide interfaces. Adv Colloid Interface Sci 1996;69:3162.
13. Matsuo T, Wise KD. Integrated field-effect electrode for biopotential recording. IEEE Trans Biomed Eng 1974;BM21:
485487.
14. Rocher V, et al. An oxynitride ISFET modified for working in a
differential-mode for pH detection. J Electrochem Soc 1994;
141:535539.
15. Wong HS, White MH. A CMOS-integrated ISFET-operational
amplifier chemical sensor employing differential sensing.
IEEE Trans Electron Devices 1989;36:479487.
16. Tsukada K, Miyahara Y, Miyagi H. Platinum-platinum oxide
gate pH ISFET. Jpn J Appl Phys Part 1-Regul Pap Short Notes
Rev Pap 1989;28:24502453.
17. Akiyama T, et al. Ion-sensitive field-effect transistors with
inorganic gate oxide for pH sensing. IEEE Trans Electron
Devices 1982;29:19361941.
18. Voigt H, et al. Diamond-like carbon-gate pH-ISFET. Sens
Actuator B-Chem 1997;44:441445.
19. Liao HK, et al. Study of amorphous tin oxide thin films for
ISFET applications. Sens Actuator B-Chem 1998;50:104
109.
20. Yin LT, et al. Study of indium tin oxide thin film for separative
extended gate ISFET. Mater Chem Phys 2001;70:1216.
21. Chin YL, et al. Titanium nitride membrane application to
extended gate field effect transistor pH sensor using VLSI
technology. Jpn J Appl Phys Part 1-Regul Pap Short Notes Rev
Pap 2001;40:63116315.
22. Katsube T, Lauks I, Zemel JN. pH-sensitive sputtered iridium
oxide-films. Sens Actuator 1982;2:399410.
23. Jakobson CG, Nemirovsky Y. 1/f noise in ion sensitive field
effect transistors from subthreshold to saturation. IEEE Trans
Electron Devices 1999;46:259261.
24. Jolly RD, McCharles RH. A low-noise amplifier for switched
capacitor filters. IEEE J Solid-State Circuit 1982;17:11921194.
25. Bousse L, Shott J, Meindl JD. A process for the combined
fabrication of ion sensors and CMOS circuits. IEEE Electron
Device Lett 1988;9:4446.
26. Bausells J, Carrabina J, Errachid A, Merlos A. Ion-sensitive
field-effect transistors fabricated in a commercial CMOS
technology. Sens Actuator B-Chem 1999;57:5662.
27. Hammond PA, Ali D, Cumming DRS. Design of a single-chip
pH sensor using a conventional 0.6 mm CMOS process. IEEE
Sens J 2004;4:706712.
28. Jakobson CG, Dinnar U, Feinsod M, Nemirovsky Y. Ionsensitive field-effect transistors in standard CMOS by post
processing. IEEE Sens J 2002;2(4):279287.
29. van der Spiegel J, Lauks I, Chan P, Babic D. The extended gate
chemically sensitive field-effect transistor as multi-species
microprobe. Sens Actuator 1983;4:291298.
30. Smith R, Huber RJ, Janata J. Electrostatically protected
ion sensitive field-effect transistors. Sens Actuator 1984;5:
127136.
31. Baldi A, Bratov A, Mas R, Domnguez C. Electrostatic discharge sensitivity tests for ISFET sensors. Sens Actuator
B-Chem 2001;80:255260.
32. Reinhoudt DN, et al. Development of durable K-selective
chemically-modified field-effect transistors with functionalized polysiloxane membranes. Anal Chem 1994;66:3618
3623.
33. Brunink JAJ, et al. Chemically modified field-effect transistors - a sodium-ion selective sensor based on calix[4]arene
receptor molecules. Anal Chim Acta 1991;254:7580.
197
34. Sibbald A, Whalley PD, Covington AK. A miniature flowthrough cell with a 4-function CHEMFET integrated-circuit
for simultaneous measurements of potassium, hy-drogen, calcium and sodium-ions. Anal Chim Acta 1984;159:4762.
35. Chovelon JM, Jaffrezic-Renault N, Fombon JJ, Pedone D.
Monitoring of ISFET encapsulation aging by impedance measurements. Sens Actuator B-Chem 1991;3:4350.
36. Grisel A, Francis C, Verney E, Mondin G. Packaging technologies for integrated electrochemical sensors. Sens Actuator
1989;17:285295.
37. Gra`cia I, Cane C, Lora-Tamayo E. Electrical characterization
of the aging of sealing materials for ISFET chemical sensors.
Sens Actuator B-Chem 1995;24:206210.
38. Mun oz J, et al. Planar compatible polymer technology for
packaging of chemical microsensors. J Electrochem Soc
1996;143:20202025.
39. Bratov A, Mun oz J, Dominguez C, Bartrol J. Photocurable
polymers applied as encapsulating materials for ISFET production. Sens Actuator B-Chem 1995;25:823825.
40. Tsukada K, Sebata M, Miyahara Y, Miyagi H. Long-life multipleISFETs with poly-meric gates. Sens Actuator 1989;18: 329336.
41. Cane C, Gra`cia I, Merlos A. Microtechnologies for pH ISFET
chemical sensors. Micro-electron J 1997;28:389405.
42. Bergveld P. The operation of an ISFET as an electronic device.
Sens Actuator 1981;1:1729.
43. Ravezzi L, Conci P. ISFET sensor coupled with CMOS readout circuit microsystem. Electron Lett 1998;34:22342235.
44. Sibbald A. A chemical-sensitive integrated-circuit - the operational transducer. Sens Actuator 1985;7:2338.
45. Pala n B, et al. New ISFET sensor interface circuit for biomedical applications. Sens Actuator B-Chem 1999;57:6368.
46. Hammond PA, Ali D, Cumming DRS. A system-on-chip digital
pH meter for use in a wireless diagnostic capsule. IEEE Trans
Biomed Eng 2005;52:687694.
47. Harame DL, Shott JD, Bousse L, Meindl JD. Implantable ionsensitive transistors. IEEE Trans Biomed Eng 1984;31:572
572.
48. Smith RL, Scott DC. An integrated sensor for electrochemical
measurements. IEEE Trans Biomed Eng 1986;33:8390.
49. Suzuki H, Shiroishi H, Sasaki S, Karube I. Microfabricated
liquid junction Ag/AgCl reference electrode and its application
to a one-chip potentiometric sensor. Anal Chem 1999;71:5069
5075.
50. Comte PA, Janata J. Field-effect transistor as a solid-state
reference electrode. Anal Chim Acta 1978;101:247252.
51. van den Berg A, Bergveld P, Reinhoudt DN, Sudho lter EJR.
Sensitivity control of ISFETs by chemical surface modification. Sens Actuator 1985;8:129148.
52. Matsuo T, Nakajima H. Characteristics of reference electrodes
using a polymer gate ISFET. Sens Actuator 1984;5:293305.
53. Bergveld P, et al. How electrical and chemical requirements
for REFETs may coincide. Sens Actuator 1989;18:309327.
54. Sudho lter EJR, et al. Modification of ISFETs by covalent
anchoring of poly(hydroxyethyl methacrylate) hydrogel-introduction of a thermodynamically defined semiconductor-sensing membrane interface. Anal Chim Acta 1990;230:5965.
55. van den Vlekkert HH, de Rooij NF, van den Berg A, Grisel A.
Multi-ion sensing system based on glass-encapsulated
ph-ISFETs and a pseudo-REFET. Sens Actuator B-Chem
1990;1:395400.
56. Chudy M, Wro blewski W, Brzo zka Z. Towards REFET. Sens
Actuator B-Chem 1999;57:4750.
57. Bergveld P. Thirty years of ISFETOLOGY-what happened in
the past 30 years and what may happen in the next 30 years.
Sens Actuator B-Chem 2003;88:120.
58. Visch LL, Bergveld P, Lamprecht W, Sgravenmade EJ. pH
measurements with an ion sensitive field-effect transistor in
198
59.
60.
61.
62.

the mouth of patients with xerostomia. IEEE Trans Biomed
Eng 1991;38:353356.
Hiraishi N, et al. Evaluation of active and arrested carious
dentin using a pH-imaging microscope and an X-ray analytical
microscope. Oper Dent 2003;28:598604.
De Aza PN, Luklinska ZB, Anseau M. Bioactivity of diopside
ceramic in human parotid saliva. J Biomed Mater Res Part B
2005;73B:5460.
Sant W, et al. Development of a creatinine-sensitive sensor for
medical analysis. Sens Actuator B-Chem 2004;103:260264.
Mackay RS. Radio telemetering from within the body. Science
1961;134:11961202.
63. Iddan G, Meron G, Glukhovsky A, Swain P. Wireless capsule

endoscopy. Nature (London) 2000;405:417417.
64. Pandolfino JE, et al. Ambulatory esophageal pH monitoring
using a wireless system. Am J Gastroenterol 2003;98:740749.
65. Johannessen EA, et al. Implementation of multichannel
sensors for remote biomedical measurements in a microsystems format. IEEE Trans Biomed Eng 2004;51:525535.
See also IMMUNOLOGICALLY
SENSITIVE FIELD-EFFECT TRANSISTORS;
INTEGRATED CIRCUIT TEMPERATURE SENSOR.
ISFET. See ION-SENSITIVE FIELD-EFFECT TRANSISTORS.
J
little movement as in the case of the rib cage that contributes to the ability of this area to expand with respiration. Most symphyses are permanent; those of sacrum and
coccyx can, however, degenerate with subsequent fusion
between adjacent vertebral bodies as part of the normal
development of these bones.
Fibrous joints are solid. The binding mechanism that
dominates the connectivity of the articulating elements is
principally fibrous connecting tissue, although other tissue
types also may be present. Length, specific arrangement,
and fiber density vary considerably according to the location of the joint and its functional requirements. Fibrous
joints are classified in three groups: sutures, gomphoses,
and syndesmoses (Table 3);
In addition to the obligatory components that all the
synovial joints possess, several joints contain intraarticular structures. Discs and menisci are examples of such
structures. They differ from one another mainly in that
a disc is a circular structure that may completely subdivide
a joint cavity so that it is, in reality, two joints in series,
whereas a meniscus is usually a crescent-shaped structure
that only partially subdivides the joint. Complete discs are
found in the sternoclavicular and in the radiocarpal joint. A
variety of functions have been proposed for intraarticular
discs and menisci. They are normally met at locations
where bone congruity is poor, and one of their main functions is to improve congruity and, therefore stability
between articular surfaces. Shock absorption facilitation
and combination of movements are among their likely
roles. They may limit a movement or distribute the weight
over a larger surface or facilitate synovial fluid circulation
throughout the joint.
The labrum is another intraarticular structure. In
humans, this structure is only found in the glenohumeral
and hip joints. They are circumferential structures
attached to the rim of the glenoid and acetabular sockets.
Labra are distinct from articular cartilage because they
consist of fibrocartilage and are triangular in their middle
section. Their bases are attached to the articular margins
and their free apical surfaces lined by synovial membrane.
Like discs, their main function is to improve fit and protect
the articular margins during extremes of movement.
Fat pads are localized accumulations of fat that are
found around several synovial joints, although only those
in the hip (acetabular pad) and the knee joint (infrapatellar
pad) are named. Suggested functions for fat pads include
protection of other intraarticular structures (e.g., the
round ligament of the head of the femur) and serving as
cushions or space-fillers thus facilitating more efficient
movement throughout the entire available range.
Bursae are enclosed, self-contained, flattened sacs typically with a synovial lining. They facilitate movement of
musculoskeletal tissues over one another and thus are
located between pairs of structures (e.g., between ligament
and tendon, two ligaments, two tendons or skin, and bone).
Deep bursae, such as the illiopsoas bursa or the deep
GEORGE PAPAIOANNOU
University of Wisconsin
YENER N. YENI
Henry Ford Hospital
Detroit, Michigan
INTRODUCTION
The human skeleton is a system of bones joined together to
form segments or links. These links are movable and
provide for the attachment of muscles, ligaments, tendons,
and so on. to produce movement. The junction of two or
more bones is called an articulation. There are a great
variety of joints even within the human body and a multitude of types among living organisms that use exo- and
endoskeletons to propel. Articulation can be classified
according to function, position, structure and degrees of
freedom for movement they allow, and so on. Joint biomechanics is a division of biomechanics that studies the effect
of forces on the joints of living organisms.
Articular Anatomy, Joint Types, and Their Function
Anatomic and structural classification of joints typically
results in three major categories, according to the predominant tissue or design supporting the articular elements
together, that is, joints are called fibrous, cartilaginous, or
synovial.
Synovial joints are cavitated. In general, two rigid
skeletal segments are brought together by a capsule of
connective tissue and several other specialized tissues,
that form a cavity. The joints of the lower and upper limbs
are mainly synovial since these are the most mobile joints.
Mobility varies considerably and a number of subcategories are defined based on the specific shape or architecture and topology of the surfaces involved (e.g., planar,
saddle, ball and socket) and on the types of movement
permitted (e.g., flexion and extension, medial and lateral
rotation) (Table 1). The basic structural characteristics
that define a synovial joint can be summarized in four
features: a fibrous capsule that forms the joint cavity, a
specialized articular cartilage covering the articular surfaces, a synovial membrane lining the inner surface of the
capsule that also secretes a special lubricating fluid, the
synovial fluid. Additional supportive structures in synovial
joints include disks, menisci, labra, fat pads, tendons, and
ligaments.
Cartilaginous joints are also solid and are more commonly known as synchondroses and symphyses, a classification based on the structural type of cartilage that
intervenes between the articulating parts (Table 2). This
cartilage is hyaline and fibrocartilage for synchondroses
and symphyses, respectively. Synchondroses allow very
199
Table 1. Diarthroses: Synovial Joints
(continued)
200
201
Table 1. (Continued )
retrocalcaneal bursa, develop along with joints and by a

similar series of events during the embryonic period.
Tendons are located at the ends of many muscles and
are the means by which these muscles are attached to bone
or other skeletal elements. The primary structural component of tendons is type I collagen. Tendons almost exclusively operate under tensile forces.
Ligaments are dense bands of connective tissue that

connect skeletal elements to each other, either creating
(as in the case of syndesmoses) or supporting joints.
According to their location they are classified as intracapsular, capsular, or extracapsular. Structurally, they
resemble the tendons in that they consist predominantly
of type I collagen.
202
Table 2. Amphiarthroses: Cartilaginous Joints
Articular Cartilage
Articular cartilage, the resilient load-bearing tissue that
forms the articulating surfaces of synovial joints functions
through load distribution mechanism by increasing the
area of contact (thereby reducing the stress) and provides
these surfaces with the low friction, lubrication, and wear
characteristics required for repetitive gliding motion.
Biomechanically, cartilage is another intraarticular
absorption mechanism that dampens mechanical shocks
and spreads the applied load onto subchondral bone
(Fig. 2). Articular cartilage should be viewed as a multiphasic material. It consists primarily of a large extracellular
matrix (ECM) with a sparse population of highly specialized
cells (chondrocytes) distributed throughout the tissue. The
primary components of the ECM are water, proteoglycans,

and collagens, with other proteins and glycoproteins present
in lower amounts (1). The solid phase is comprised by this
porous-permeable collagen-PG matrix filed with freely
movable interstitial fluid (fluid phase) (2). A third phase
is the ion phase, necessary to describe the electromechanical
behaviors of the system. The structure and composition of
the articular cartilage vary throughout its depth (Fig. 2),
from the articular surface to the subchondral bone. These
differences include cell shape and volume, collagen fibril
diameter and orientation, proteoglycan concentration, and
water content. These all combine to provide the tissue with
its unique and complex structure and mechanical properties. A fine mechanism of interstitial fluid pressurization
Table 3. Synarthroses: Fibrous Joints
Figure 1. Basic structure and components of a synovial joint (also

called diarthroses).
Figure 2. Zones of articular cartilage.
203
204
results from the flow of interstitial fluid through the porouspermeable solid matrix that in turn defines the rate dependent load-bearing response of the material. It is noteworthy
that articular cartilage provides its essential biomechanical
functions for eight decades or more in most of the human
synovial joints and no synthetic material performs this well
as a joint surface.
The frictional characteristics between two surfaces sliding over each other are significantly influenced by the
topography of the given surfaces. Anatomical shape
changes affect the way in which loads are transmitted
across joints, altering the lubrication mode in that joint
and, thus, the physiologic state of cartilage. Articular
surfaces are relatively rough, compared to machined bearing surfaces, at the microscopic level. The natural surfaces
are surprisingly much rougher than joint replacement
prostheses. The mean of the surface roughness for articular
cartilage ranges from 1 to 6 mm, while the metal femoral
head of a typical artificial hip has a value of 0.025 mm,
indicating that the femoral head is apparently much
smoother. Topographic features on the joint surfaces are
characterized normally by primary anatomic contours,
secondary roughness (<0.5 mm in diameter and <50 mm
deep), tertiary hollows on the order of 2045 mm deep; and,
finally, quaternary ridges 14 mm in diameter and 0.10.3
mm deep. Scanning electron micrographs (SEMs)of
arthritic cartilage usually depict a large degree of surface
irregularity and anomalous microtopography. These surface irregularities have profound effects on the lubrication
mechanism. They accelerate the effects of friction and the
rate of degradation of the articular cartilage. The types of
joint surface interactions vary greatly between different
joints in the body, different animals, between different size
animals of the same species, different genders, and different ages. For example, the human hip joint is a deep
congruent ball and socket joint (where the cartilage thickens peripherally at the acetabulum); this differs greatly
from the bcondylar nature of the distal femur in the knee
joint, and the saddle shape of the carpometacarpal joint in
the thumb. The degree of shape matching between the
various bones and articulating cartilage surfaces composing a joint is a major factor affecting the distribution of
stresses in the cartilage and subchondral bone.
Effects of Motion and External Loading on Joints
The articular joint is viewed as an organ with complicated
mechanisms of memory and adaptation that accommodates changes in its function. Joint loading results in
motion and the couple loadmotion is required to maintain
normal adult articular cartilage composition, structure,
and mechanical properties. The type, intensity, and frequency of loading necessary to maintain normal articular
cartilage vary over a broad range. The intensity or frequency of loading should not exceed or fall below these
necessary levels, since this will disturb the balance
between the processes of synthesis and degradation.
Changes in the composition and microstructure of cartilage
will result. Reduced joint loading, as has been observed in
cases of immobilization by casting or external fixation,
leads to atrophy or degeneration of the cartilage. The
changes affect both the contact and noncontact areas.

Changes in the noncontact areas resulting from rigid
immobilization include fibrillation, decreased proteoglycan
content and synthesis, and altered proteoglycan conformation, such as a decrease in the size of aggregates and amount
of aggregate. Normal nutritive transport to cartilage from
the synovial fluid by means of diffusion and convection has
been diminished, resulting in these changes. Increased joint
loading, either through excessive use, increased magnitudes
of loading, or impact, also may affect articular cartilage.
Catabolic effects can be induced by a single-impact or
repetitive trauma, and may serve as the initiating factor
for progressive degenerative changes. Osteoarthritis, a joint
disease of epidemic proportions in the western world, is
characterized by erosive cartilage lesions, cartilage loss and
destruction, subchondral bone sclerosis and cysts, and large
osteophyte formation at the margins of the joint (3).
Moderate running exercise may increase the articular
cartilage proteoglycan content and compressive stiffness,
decrease the rate of fluid flux during loading, and increase
the articular cartilage thickness in skeletally immature
animals. However, no significant changes in articular cartilage mechanical properties were observed in dogs in
response to lifelong increased activity that did not include
high impact or torsional loading of their joints. Disruption
of the intraarticular structures (e.g., menisci or ligaments)
will alter the forces acting on the articular surface in both
magnitude and areas of loading. The resulting joint
instability is associated with profound and progressive
changes in the biochemical composition and mechanical
properties of articular cartilage. In experimental animal
models, responses to transection of the anterior cruciate
ligament or meniscectomy have included fibrillation of the
cartilage surface, increased hydration, changes in the proteoglycan content, reduced number and size of proteoglycan aggregates, joint capsule thickening, and osteophyte
formation. It seems likely that some of these changes result
from the activities of the chondrocytes, because their rates
of synthesis of matrix components, breakdown of matrix
components, and secretion of proteolytic enzymes are all
increased. In vitro studies have shown that loading of the
cartilage matrix can cause all of these mechanical, electric,
and physicochemical events, but thus far it has not been
clearly demonstrated which signals are most important in
stimulating the anabolic and catabolic activity of the chondrocytes. A holistic physicochemical and biomechanical
model of cartilage function in health and disease remains
a challenge in the scientific community.
KINEMATICS OF JOINTS
General Comments
Mechanical analysis can refer to kinetics (forces) and/or
kinematics (movement), with kinetics being the cause and
kinematics the result. Mechanical analysis can develop
models proceeding from forces to movements or vice versa.
The analysis that starts from the cause (force) is called
direct or forward dynamics, and produces a defined set of
forces that caused the unique movement. This approach
has one solution, and hence is deterministic. Starting from
the movement the analysis is called inverse dynamics. In

this case, an infinite number of combinations of individual
forces acting on the system can be the causes of the same
unique movement, which makes the inverse dynamics
approach not deterministic. The simplest and most essential system of mechanical formulations for explaining and
describing motion is the Newtons second law. More
advanced techniques include the Lagrange, dAlembert,
and Hamiltons methods. In general all of these methods
start by describing equations of motion for a rigid body for
translation, rotation, or combinations of them for both two
(2D) and three-dimensional (3D) space. If the model
assumes that the articulated segments that create the
articulation are modeled as rigid bodies the remaining task
is to calculate the relative motion between the two segments by applying graphics or joint kinematic analysis.
Kinematics is the study of the movements of rigid
structures, independent of the forces that might be
involved. Two types of movement, translation (linear displacement) and rotation (angular displacement), occur
within three orthogonal planes; that is, movement has
six degrees of freedom. Humans belong to the vertebrate
portion of the phylum chordata, and as such possess a bony
endoskeleton that includes a segmented spine and paired
extremities. Each extremity is composed of articulated
skeletal segments linked together by connective tissue
elements and surrounded by skeletal muscle. Motion
between skeletal segments occurs at joints. Most joint
motion is minimally translational and primarily rotational.
The deviation from absolute rotatory motion may be noted
by the changes in the path of a joints instantaneous center
of rotation. These paths have been measured for most of
the joints in the body and vary only slightly from true arcs
of rotation. For human motion to be effective, not only must
a comparatively rigid segment rotate its position relative to
an adjacent segment, but many adjacent limb movements
must interact as well. Whether the hand is trying to write
or the foot must be lifted high enough to clear an obstacle on
the ground, the activity is achieved via coordinated movements of multiple limb segments. To provide for the greatest
possible function of an extremity, the proximal joint must
have the widest range of motion to position the limb in space.
This joint must allow for rotatory motions of large degrees
in all three planes about all three axes. A means is also
provided to translate the limb, so that an extremity can
function at all locations within its global range. Rotational
motion of the elbow and knee joints allows such overall
changes as adjacent limb segments move. Finally, to finetune the use of this mechanism with respect to the extremities, for their functional purposes, the hand and foot
are required to have a vast amount of movement about all
three axes, although the rigid segments are relatively small.
Such movement requires the presence of relatively universal joints at the terminal aspect of each extremity.
Characterization of the General Mechanical Joint System:
Terminology and Definitions
The displacement of a point is simply the difference
between its position after a motion and its position before
that motion. It can be represented by a 3D vector drawn
205
from the initial position of the point to its final position. The
components of the displacement vector will be the changes
in the coordinates of the points position from measurement
in the reference coordinate system. It is apparent that not
only the positions, but also displacements measured are
relative to some reference. Rigid body (RB) displacements
are more complicated than point displacements since for a
rigid body a displacement is a change in its position relative
to some reference, but more than three parameters are
needed to describe it. Two simple types of RB displacement
can be described: translation and rotation. An important
property of pure translation of a RB is that the displacement
vectors of all points in the body are identical and are nonzero.
In pure rotation of a RB, although points in the body experience nonzero displacements, one point in that body experiences zero displacement. In addition to that rule, Eulers
theorem shows that in pure rotation all points along a
particular line through that undisplaced point also experience zero displacement. This line is also known as the axis
of rotational displacement. Chasles theorem further states
that any displacement of a RB can be accomplished by a
translation along a line parallel to the axis of rotation that is
defined by Eulers theorem plus a rotation about that same
parallel axis. Simply that suggests that any displacement in
3D is equivalent to the motion of a nut, representing the body,
on an appropriate stationary screw that was centered on the
line described above. Indeed, it can be shown that any displacement in 3D is equivalent to a translation plus a rotation.
Degrees of Freedom
The biological organisms capable of propelling themselves
through different media consist of more than one rigid
body. A system consisting of a 3D reference frame and
an isolated rigid body in space has six degrees of freedom
(DOF). To describe the position of each body relative to the
ground reference frame it would be necessary to use six
parameters, so for two unconnected rigid bodies 12 parameters would be necessary. The system consisting of these
two unconnected bodies and the fixed -ground reference
would have 12 DOF. The humananimal body consists of a
combination of suitably connected bodies. The connections,
joints between the bodies, serve to constrain the motions of
the bodies so that they are not free to move with what
would otherwise be six DOF for each body. Therefore, we
can define the number of DOF that the joint removes as
the number of degrees of constraint that it provides. It can
be shown that every time a joint is added to a system, the
number of degrees of freedom in that system is reduced by
the number of degrees of constraint provided by that joint.
This suggests the following generic formula for the calculation of the degrees of freedom of a system:
F 6L 1 5J1 4J2 3J3 2J4 J5
where F is the number of degrees of freedom in the
system of connected joints; L is the number of joints in
the system, including the ground joint (which has no
degrees of freedom), and Jn the number of joints having
n degrees of freedom each.
Table 4 contains a description of the major joints in the
human body along with the segments-bones that they
206
Table 4. Characteristics of Major Human Joints

Joint
Bones
Type
DOF
Shoulder
Humerus-Scapula
Diarthrosis (spheroidal)
Elbow
Humerus-ulna
Diarthrosis (ginglymus)
Radioulnar
Superior radius-ulna
Diarthrosis (trochoid)
Wrist
Radius-carpal
Diarthrosis (condyloid)
Metacarpal-phalangeal
Metacarpal-phalanges
Finger
Interphalanges
Thumb
First metacarpal-carpal
Diarthrosis (reciprocal)
Hip
Femur-acetabulum
Diarthrosis (spheroidal)
Knee
Tibia-femur
Ankle
Tibia-fibula-talus
Intertarsal
Metatarsal-phalangeal
Tarsals
Metatarsals-phalanges
Diarthrosis (arthroidal)
2
2
Interphalangeal
Phalanges
Tibio-fibular
Skull
Sterno-costal
Sacroiliac
Intervertebral
Distal tibia-fibula
Cranial
Ribs-sternum
Sacrum-ilium
Cervical vertebrae
Synarthrosis (syndesmosis)
Synarthrosis (suture)
Amphiarthrosis (synchondrosis)
Amphiarthrosis (synchondrosis)
0
0
0
0
3
Thoracic vertebrae
Lumbar vertebrae
Atlas axis
Diarthrosis (trochoid)
Type of motion
Range of Motion (deg)
Flexion
Extension
Abduction
Abduction
Rotation
Flexion
Extension
Rotation (radius)
Pronation
Supination
Flexion
Extension
Radial deviation
Ulnar deviation
Circumduction
Flexion
Extension
Radial deviation
Ulnar deviation
Flexion
Extension
Flexion
Extension
Abduction
Abduction
Circumduction
Flexion
Extension
Abduction
Abduction
Medial rotation
Lateral rotation
Circumduction
Flexion
Extension
Medial rotation
Lateral rotation
Flexion
Extension
Gliding
Flexion
Extension
Abduction
Abduction
Flexion
Extension
Slight movement
No movement
Slight movement
No movement
Flexion
Extension
Lateral flexion
Axial rotation
Flexion
Extension
Lateral flexion
Axial rotation
Flexion
Extension
Lateral flexion
Axial rotation
Pivoting motion
150
5060
90120
Complete
145160
05
7075
8590
9095
6070
2025
5565
Complete
8090
2030
2025
1520
8090
010
8090
2025
4045
010
Complete
90120
1020
3045
30
3040
60
Complete
120140
0
30
40
2030
4045
Limited motion
2530
8090
1520
Limited
90
0
Give
Elastic
40
75
3545
4550
105
60
20
35
60
35
20
5
articulate, their respective type DOF and type/range of

motion they provide.
Planar Motion
Some human joints move predominantly in one plane (e.g.,
the knee joint) in which case the motion can be approximated and analyzed by graphical methods. Here the rotation is characterized by the motion of all points on
concentric cycles with an identical angle of rotation around
the undisplaced center of rotation (CR). The CR may be
located inside and outside of the boundaries of the rotating
body. The most common graphical method for the calculation is the so-called bisection technique. If the initial and
final states of the body are known, the position of the center
or rotation and the angle of rotation may be reconstructed
(Fig. 3).
Instantaneous Center of Rotation
When a 2D body is rotating without translation, for example, a rotating stationary bicycle gear, any marked point P
on the body may be observed to move in a circle about a
fixed point called the axis of rotation or center of rotation.
When a rigid body is both rotating and translating, for
example, the motion of the femur during gait, its motion at
any instant of time, can be described as rotation around a
moving center of rotation. The location of this point at any
instant, termed the instantaneous center of rotation (ICR),
is determined by finding the point which, at that instant, is
not translating. Then by definition, at that instant, all
points on the rigid body are rotating about the ICR. For
practical purposes, the ICR is determined by noting the
paths traveled by two points, S and Q, on the object in a

0
0
0
0
very short period of time to S and Q . The paths SS and QQ
will be perpendicular to lines connecting them to the ICR
because they approximate, over short periods, tangents to
the circles describing the rotation of the body around the
ICR at that instant. Perpendicular bisectors to these two
paths will intersect at the instantaneous axis of rotation.
If the ICR is considered to be a point on a moving body,
its path on the fixed body is called a fixed centrode. If the
ICR is considered to be a point on a fixed body, its path on
the moving body is called the moving centrode.
Although in principle two objects may move relative to
one another in any combination of rotation and translation,
diarthroidal joint surfaces are constrained in their relative
motion. The articular surface geometry, the ligamentous
restraints, and the action of muscles spanning the joint are
the main constraining systems. In general, joint surface
separation (or gapping-proximal-distal) and impaction are
small compared to overall joint motion. Mechanically,
when surfaces are adjacent to each other they may move
relative to each other in either sliding or rolling contact. In
rolling contact, the contacting points on the two surfaces
have zero relative velocity, that is, no slip. Rolling and
sliding contacts occur together when the relative velocity at
the contact point is not zero. The instant center will then lie
between the geometric center and the contact point. All
diarthroidal joint motion consists of both rolling and sliding motion. In the hip and shoulder, sliding motion predominates over rolling motion. In the knee, both rolling
and sliding articulation occur simultaneously. These simple concepts affect the design of total joint prostheses. For
example, some total knee replacements have been designed
for implantation while preserving the posterior cruciate
ligament, which appears to help maintain the normal
kinematics of rolling and sliding in the knee. Other knee
prostheses substitute for ligament control of kinematics by
alterations in articular surface contour through constraining congruity.
Analytical Methods
ICR
Q
S
207
Figure 3. Points S and S0 as well as Q and Q0 , lie on the arcs of

circles around the center of rotation ICR (used synonymously with
CR after the section Instantaneous Center of Rotation). If lines
SS0 and QQ0 are bisected perpendicularly, the center of rotation
CR is located at the intersection of these perpendicular bisectors.
This construction assumes that the perpendicular bisectors are differently orientated but a special case arises if the bisectors are
identically oriented. Then the points S, Q and the center of rotation
ICR lie on a straight line.
Simple kinematic analysis of pure planar translations and

rotation or combinations of the two as well as complicated
3D analysis of a rigid body requires the positional information of a minimum of three noncolinear points to describe
this motion uniquely. If the position of three points at
two instants is known, the displacement from one position to another may be interpreted as translation, rotation
or both. Therefore, the first task is to continually monitor
the positions of three points on each rigid body. This analysis is conveniently divided into data collection and data
analysis.
Data Collection. A constant challenge for the experimental motion analyst is the collection of accurate spatial
displacement kinematics of a joint. Several methods have
been employed. A review is presented here.
Video and digital optical motion capture (tracking)
systems offer state-of-the-art, high resolution, accurate
motion capture options to acquire, analyze, and display
3D motion data. The systems are integrated with analog
208
data acquisition systems to enable simultaneous acquisition (1300 Hz) of force plate and electromyographic data.
Clinically validated software analysis packages are used to
analyze and display kinematic, kinetic, and electromyographic data in forms that are easy to interpret.
The major components of a video and digital motion
capture system are the cameras, the controlling hardware
modules, the software to analyze and present the data, and
the host computer to run the software. These systems are
designed to be flexible, expandable (from 3 to up to 200
cameras in motion analysis tracking for Hollywood animation
movies) and easy to integrate into any working environment.
This system collects and processes coordinate data in the
least amount of time and requires minimal operator intervention. This system uses motion capture cameras to rapidly
acquire 3D coordinate positions from reflective markers
placed on subjects. Illuminating strobes with differing wavelengths are used to track the spatial displacement (between 1
and 10 mm resolution) of spherical reflective markers
attached to the subjects skin at appropriately chosen locations, preferably on bony landmarks on the human body to
minimize skin movement. They can be infrared (IR), visible
red, or near-IR strobes to fit the lighting conditions of the
capture environment. Also, the lenses can be of fixed or
variable focal length for total adaptability. Images are
processed within the optical capture cameras where markers are identified and coordinates are calculated before
being transferred to the computers. After the completion
of the movement, the system provides 3D coordinate and
kinematic data. The disadvantages of the system include
the skin movement error whose effect is more prominent
(3 cm error) at high movement speeds. These high speed
motion tasks (impact biomechanics, e.g.) are handled by
high speed cine cameras with data acquisition rates several orders of magnitude greater than clinical motion
analysis systems. The processing method that is almost
real time uses combinations of skin markers (minimum
three at each segment) to produce coordinate systems for
each segment and eventually describe intersegmental
relative motion or relate all the different segment motions
to the laboratory fixed-coordinate system. Recently, methods
employing clusters of markers have shown to somewhat
reduce the skin marker artifact but are yet to be adopted
in the clinical practice.
A more accurate method (<1 mm translation and up to
1000 Hz) is the cineradiographical method, which employs
an X-ray machine and uses special cameras for capture of
sequences of the digital radiographs. In addition to accuracy, these systems directly access the in vivo skeletal
kinematics so that the resulting analysis can be directly
related to bony landmarks. Radiation issues, magnification, and distortion factors are some drawbacks that can be
overcome by appropriate image analysis techniques. This
method is, however, prone to occlusion errors when two
segments overlap and simultaneously cross the field of
view of the X-ray source. Stereosystems with more than
one X-ray sources can limit this artifact. A biplane radiographic system consists of two X-ray generators and two
image intensifiers optically coupled to synchronized high
speed video cameras that can be configured in a custom
gantry to enable a variety of motion studies. The system
can be set up with various set-up modes (e.g., a 608 interbeam angle), an X-ray source to object distance of 1.3 m,
and an object to intensifier distance of 0.5 m. Images are
acquired with the generators in continuous radiographic
mode (typically 100 mA, 90 kVp). The video cameras are
electronically shuttered to reduce motion blur. Short (0.5 s)
sequences are recorded to minimize radiation exposure.
X-ray exposure and image acquisition are controlled by
an electronic timersequencer to capture only the desired
phase of movement.
CODA is an acronym of Cartesian Opto-electronic
Dynamic Anthropometer, a name first coined in 1974 to
give a working title to an early research instrument developed at Loughborough University, United Kingdom. The
3D capability is an intrinsic characteristic of the design of
the sensor units, equivalent to but much more accurate
than the stereoscopic depth perception in normal human
vision. The system is precalibrated for 3D measurement,
which means that the lightweight sensor can be set up at a
new location in a matter of minutes, without the need to
recalibrate using a space-frame. Each sensor unit must be
independently capable of measuring the 3D coordinates of
skin markers in real-time. As a consequence, there is great
flexibility in the way the system can be operated. For
example, a single sensor unit can be used to acquire 3D
data in a wide variety of projects, such as unilateral gait. Up
to six sensor units can be used together and placed around a
capture volume to give extra sets of eyes and maximum
redundancy of viewpoint. This enables the system to track
3608 movements that often occur in animation and sports
applications. The calculation of the 3D coordinates of markers is done in real-time with an extremely low delay of 5
ms. Special versions of the system are available with latency
shorter than 1 ms. This opens up many applications that
require real-time feedback, such as research in neurophysiology and high quality virtual reality systems, as well
as tightly coupled real-time animation. It is also possible to
trigger external equipment using the real-time data. The
automatic intrinsic identification of markers combined with
processing of all 3D coordinates in real-time means that
graphs and stick figures of the motion and many types of
calculated data can be displayed on a computer screen
during and immediately after the movement occurs. The
data are also immediately stored to file on the hard drive.
A new concept in measuring movement disorders utilizes a unique miniature solid-state gyroscope, not to be
confused with gravity sensitive accelerometers. The instrument is fixed with straps directly on the skin surface of the
structure whose motion is of interest. It has been successfully used to quantify: tremor (resting, posture, kinetic),
rapid pronationsupination of the hand, arm swing, lateral
truncal sway, leg stride, spasticity (pendulum drop test),
dyskinesia, and alternating dystonia. The system (Motus)
senses rotational motion only and is ideal for quantifying
human movement since most skeletal joints produce rotational motion. This disadvantage is outweighed by its
miniature size that allows it to be of great value for certain
types of studies. A different system (Gypsy Gyro) uses 18
small solid-state inertial sensors (gyros) to accurately
measure the exact rotations of the actors bones in real-time
for motion capture. The system can easily be worn beneath
normal clothing. With wireless range these systemssuits

can be used to record up to 64 actors simultaneously.
Another concept for 3D motion analysis is the measurement system CMS10 (Zebris) designed as a compact device
for everyday use. The measurement procedure is based on
the travel time measurement of ultrasonic pulses that are
emitted by miniature transmitters (markers placed on the
skin) to the three microphones built into the compact
device. A socket for the power pack (supplied with the
device) as well as the interface to a computer are located
on the back of the device. The evaluation and display of the
measurement data are carried out in real-time. It is possible to use either a table clamp or a mobile floor stand with
two joints to support the measurement system.
Data Analysis
Coordinate Systems and Transformation. In the analysis
of experimental joint mechanics data, the transformation
of point coordinates from one coordinate system to another
is a frequent task (4). A typical application of such a
transformation would be gait analysis data recorded in a
laboratory fixed coordinate system (by means of film or
video sequences) that must be converted to a reference
system fixed to the skeleton of the test subject. The laboratory fixed coordinate system may be designated by xyz and
the body reference system by abc (Fig. 4). The location of a
point S(a/b/c) in the body reference system is defined by
the radius vector s a ea b eb c ec . Consider the
reference system to be embedded into the laboratory system. Then the radius vector rm xm ex ym ey zm ez
describes the origin of the reference system in the laboratory system. The location of S(x/y/z) is now expressed by
the coordinates a, b, c. The vector equation r rm s gives
the radius vector for point S in the laboratory system (Fig. 4).
Employing the full notation we have: r x ex y ey
z ez xm ex ym ey zm ez a ea b eb c ec . A
set of transformation equations results after some intermediate matrix algebra to describe the coordinates. The
scalar products of the unit vectors in the xyz and abc
ec
S
eb
S
ez
rm
ea
ey
ex
Figure 4. Changing the coordinate systems, transformation of
point coordinates from one coordinate system to another.
209
t
R
t
Q
t
Figure 5. A rigid body (shoebox) moves parallel to itself. The

radius vectors from O to P and from O to P0 are designated by r and
r0 so that r0 r t, where t is the difference vector.
systems produce a set of nine coefficients Cij. The cosine

of the angle between the coordinate axes of the two
systems corresponds to the value of the scalar products.
Three direction cosines define the orientation of each
unit vector in one system with respect to the three unit
vectors of the other system. Due to the inherent properties
of orthogonality and unit length of the unit vectors, there
are six constraints on the nine direction cosines, which
leaves only three independent parameters describing the
transformation. Employing the matrix notation of the
transformation equation we have
2 3 2
3 2
3 2 3
xm
x
c11 c12 c13
a
6 7 6
7 6
7 6 7
4 y 5 4 ym 5 4 c21 c22 c23 5 4 b 5
z
zm
c31 c32 c33
In coordinate transformations, the objects remain

unchanged and only their location and orientation are
described in a rotated and possibly translated coordinate
system. If a measurement provides the relative spatial location and orientation of two-coordinate systems the relative
translation of the two systems and the nine coefficients Cij
can be calculated. The coefficients are adequate to describe
the relative rotation between the two coordinate systems.
Translation in Three-Dimensional Space. In translation
in 3D space, the rigid object moves parallel to itself (Fig. 5).
Pure translation in 3D space leaves the orientation of the
body unchanged as in the case of pure 2D translation.
Rotations about the Coordinate Axes. A rotation in 3D
space is defined by specifying an axis and an angle of
rotation (Fig. 6). The axis can be described by its 3D
orientation and location (5). A rotation, as does the translation explained earlier, leaves all the points on the axis
unchanged; all other points move along circular arcs in
planes oriented perpendicular to the axis (6,7).
This rotation moves an arbitrary point P to location P0
with constant distance z from the xy plane (z z0 ). This
produces the following matrix notation for the respective
equations for the rotation that changes x and y coordinates
210
Matrix intermediate calculation here gives

r00 Dz0 Dx0 r
R
P (x /y /z )
P (x /y /z )
r
r00 Dx Dz0 r
r
ez
O
ey
ex
y
Figure 6. Rotation about the z axis of the coordinate system.
but leaves the z coordinate unchanged.

2 03 2
32 3
cosg sing 0
x
x
6
7
6
76 7
0
0
r 4 y 5 4 sing cosg
0 54 y 5 Dz gr
0
0
1
z
z0
The matrix describing a rotation about the z axis is designated Dz(l). The matrices describing a rotation about the y axis
through angle b and about x axis through angle a are similar.
2 03 2
3 2 3
cosb
0 sinb
x
x
6 7 6
7 6 7
r0 4 y0 5 4 0
1 0

5 4 y 5 Dy br
0
sinb 0 cosb
z
z
3 2
1
x0
6 7 6
r0 4 y0 5 4 0
0
z0
2
0
cosa
sina
For rotations about body-fixed axes it is true that in general, the matrix of the last rotation in the sequence of
rotations is the first one to be multiplied by the vector to
be rotated. The matrix B describing the image resulting
from n partial rotations about body-fixed axes is composed
according to the formula:
x
x
In this calculation the sequence of the matrices is very

important especially as this sequence differs from what one
might expect. First, the matrix of the second partial rotation acts on the vector r and then, in a second step on the
matrix of the first partial rotation. If the sequence of the
two partial rotations is interchanged, the combined rotation is described by
32 3
x
76 7
sina 54 y 5 Dy ar
z
cosa
0
Combined Rotations as a Result of a Sequence of Rotations.

Assume that the first rotation of a rigid body occurs about
the z axis of a coordinate system. The rotation matrix
related to the unit vectors ex, ey, ez, is
3
2
0 1 0
7
6
05
Dz g 90
4 1 0
0 0
1
The second rotation occurs supposedly about the x0 axis,
that is, about a body-fixed axis on the body (previously
rotated about its z axis). The rotation matrix related to the
unit vectors e0x , e0y , e0z , is
2
3
1 0 0
6
7
Dx0 a 90
4 0 0 1 5
0 1 0
Bbody - fixed D1 D2 D3 . . . Dn1 Dn

where the indexes indicate the sequence of the rotations.
Alternatively, if the n rotation were to be produced about
axes fixed in space (i.e., fixed in the ground, laboratory
frame) and not about body-fixed axes, the sequence of the
matrexes in the matrix product would be different:
Bspace - fixed Dn Dn1 . . . D2 D1
Euler and BryantCardan Angles. Any desired orientation of a body can be obtained by performing rotations
about three axes in sequence. There are, however, many
ways of performing three such rotations. One can do this
task at random, but for reasons of clarity two conventions
are frequently used: the Eulers and BryantCardans
rotations. In the Euler notation, the general rotation is
decomposed of three rotations about body-fixed axes in the
following manner:
Rotation 1: about the z axis through the angle w rotation
matrix Dz(w) (Fig. 7).
0
Rotation 2: about the x axis through the angle u rotation
matrix Dx0 u.
00
Rotation 3: about the z axis through the angle c rotation matrix Dz00 c.
The matrix describing Eulers combined rotation is given by

the matrix product
B Dz w Dx0 u Dz00 c Euler
According to the Bryant and Cardan the general rotation is
decomposed of three rotations about body-fixed axes in the
following manner:
Rotation 1: about the x axis through the angle w1 rotation matrix Dx(w1) (Fig. 7).
0
Rotation 2: about the y axis through the angle w2
rotation matrix Dy0 w2 .
00
Rotation 3: about the z axis through the angle w3
rotation matrix Dz00 w3 ,
211
vides insights into the cause of the observed motion, it is

essential to the proper interpretation of human movement
processes. Forces and loads are not visually observable;
they must be either measured with instrumentation or
calculated from kinematics data. Kinetic quantities studied include such parameters as the forces produced by
muscles; reaction loads between body parts as well as their
interactions with external surfaces; the load transmitted
through the joints; the power transferred between body
segments; and the mechanical energy of body segments.
Inherent to such studies are the functional demands
imposed on the body. The structure and stability of each
extremity and its joints reflect different systems and functional demands. The functional demands on the upper
extremity are quite different from those on either the upper
and lower axial skeleton or those on the lower extremity.
Depending on which joint and/or structure is addressed,
different types and degrees of rotational motion are
allowed and are functional. How much structural strength
is needed versus how much movement is allowed in each
area dictates the nature of the material, size, shape, and
infrastructure of the joint system established to perform a
given movement.
z
z
j1
x = x
y
z = z
Equations of Motion
x
j
y
x
y
Figure 7. General rotation composed of three partial rotations.

The first rotation according to the BryantCardan convention
(above). The first of the general rotations using Euler as the
selection of the axes and angles of rotation (below).
in which case the matrix of combined rotation is given

by
B Dx w1 Dy0 w2 Dz00 w3 Bryant Cardan
For reasons of simplicity, we have presented single or
combined rotations about coordinate axes, but more
complicated rotational laws can be applied as we deal with
rotations about arbitrary axes. Rotation and translation
can also be integrated into one single motion with Chasles
theorem. Chasles theorem states that the general
motion in 3D space is helical motion, or the basic type
of motion adapted to describe any change of location and
orientation in 3D space is helical motion. The relevant
axis of rotation is designated the helical axis. Chasles
theorem is also known as the helical axis theorem.
KINETICS OF JOINTS
The study of the forces that bring about the movements
discussed above is called kinetics. Because kinetics pro-
The kinetics deal with the effects of forces on the motion of

a body. When the motion is known, the problem is then to
find the force system acting on the body. There are joint
forces and joint moments. With all the kinematic quantities
known, it is possible to find the joint forces and moments
from the resulting force system that acts on each element.
This is done by solving a system of simultaneous equations
at successive time intervals. Since muscles are an
unknown force system, the resolved muscle force and the
real joint force are treated as totally unknown joint forces
in the analysis. The three equations of motion for linear
motions are
X
X
X
F max
F may
F maz
The three equations of motion for rotation are
X
Mx Ixx ax Iyy Izz vy vz Ixy ay vx vz
Iyz v2y v2z Ixx az vx vy
X
My Iyy ay Izz Ixx vz vx Iyz az vy vx
Ixx v2z v2x Ixy ax vy vz
X
Mz Izz az Ixx Iyy vx vy Izx ax vz vy
Ixy v2x v2y Iyz ay vz vx
where M is the moment, I is the mass moment of inertia, a
the angular acceleration, and v is the angular velocity. The
moment equations can be simplified if the axes of the
reference frames coincide with the principal axes, with
the origin at the center of gravity. These equations, called
Euler equations, are
X
Mx Ix ax Iy Iz vy vz
X
My Iy ay Iz Ix vz vx
X
Mz Iz az Ix Iy vx vy
212
Continuity conditions are derived based on the fact that

equal and opposite forces and moments occur at the joint
between the two segments.
The anthropometric data for the mass, the center of
gravity, the moment of inertia, and so on for the different
parts of the human body are available in the literature
(8,9).
Motion and Forces on Diarthroidal Joints
In vivo experimental measurements on the relative motions
between articulating surfaces of a joint, which correspond to
daily activities, are limited. Most quantitative information
is obtained from gait studies that do not provide the accuracy and precision for the detailed information required for
lubrication studies. However, even simple calculations show
that translational speeds between two articulating surfaces
can range from 0.06 m.s1 between the femoral head
surface and the acetabulum surface during normal walking,
to 0.6 m.s1 between the humeral head surface and the
glenoid surface of the shoulder when a baseball pitcher
throws a fastball. Cartilage to cartilage contact or fluid-film
layers, or a mixture of both are normally the contact
mechanisms at the joint. During a normal walking cycle,
the human hip, knee, and ankle joints can be subjected to
loads on the order of six times body weight, with these peak
loads occurring just after heel-strike and just before toe-off.
The average load on the joint is approximately three to five
times body weight, which lasts as long as 60% of the walking
cycle. During the swing phase of walking, only light loads
are carried. During this phase, the articular surfaces move
rapidly over each other. In addition, extremely high forces
occur across the joints in the leg during jumping. Descending stairs can load the knee with up to 10 times body weight,
suggesting that the load on the joint surface is dependent on
the task performed, that is, the loading sites change drastically as the articulating surfaces move relative to each other.
dynamic self-stabilization from a feedforward, tuned

mechanical system can reject rapid perturbations and
simplify control. Future progress would benefit from the
creation of a field embracing comparative neuromechanics.
Both templates and anchors are part of a system of mathematical and structural definitions and standard methods
of description and dissemination of knowledge. In the next
few sessions an attempt is made to describe some of those
methods.
The human musculoskeletal system is often modeled by
joining rigid links with continuous mass distribution. The
joint may be of the revolute or spherical type, with restrictions consistent with body construction. Usually, the
human segments form an open linkage.
Knowing all the kinematics at the center of mass of the
segments, the joint force and the moment analysis proceeds
by drawing free-body diagrams of the segments involved.
The free body diagrams for the hip joint, the knee joint,
and the ankle joint in a sagittal plane are illustrated in
Fig. 8 as an example:
Assessment of Mechanical Factors Associated With Joint
Degeneration: Limitations and Future Work
Joint degeneration results from complex, multidimensional, nonlinear, dynamically coupled interactions
between the organism and its environment. The assessment of mechanical factors associated with joint degeneration has traditionally combined longitudinal clinical
studies with carefully designed experimental techniques
and theoretical computational analyses. The quality of
MATHEMATICAL AND MECHANICAL MODELS OF JOINTS

Locomotion results from complex, high dimensional, nonlinear, dynamically coupled interactions between an
organism and its environment. Simple models called templates have been and can be made to resolve the redundancy of multiple legs, joints, and muscles by seeking
synergies and symmetries. The simplest model (least number of variables and parameters) that exhibits a targeted
behavior is called a template (10). Templates can be used to
test control strategies against empirical data. Templates
must be based in more detailed morphological and physiological models to ask specific questions about multiple legs,
the joint torques that actuate them, the recruitment of
muscles that produce those torques and the neural networks that activate the ensemble. These more elaborate
models are called anchors. They introduce representations
of specific biological details of the organism. The control of
slow, variable-frequency locomotion appears to be dominated by the nervous system, whereas during rapid, rhythmic locomotion, the control may reside more within the
mechanical system. Anchored templates of many-legged,
sprawled-postured animals may suggest that passive,
Figure 8. Relationship between the free body diagram and the

link-segment model. Each segment is broken at the joints, and
the reaction forces and moments of force acting at each joint are
shown.
such assessments depends both on the accuracyprecision

of the measurement methodology and the theoretical framework for its interpretation (i.e. joint mathematical models). To capitalize on the increasing level of measurement
accuracy, theoretical analysis requires more detailed morphological and physiological models (11). For example,
because of variations between individuals, the detailed,
morphological analysis required for accurate modeling of
cartilage stresses must be patient specific. In addition, the
dynamics of the human task to be modeled (for e.g., human
jumping) as expressed in the strain rate of tissue deformation must be accounted for in the analysis. Once the error
estimate (simulation versus experiment) is established,
model predictions can address specific clinical-biological
questions. Traditionally, in situ methodology (cadaveric
experiments and in vitro tests) is applied when an in vivo
measurement is impossible. This limitation presents a
number of implications and assumptions that weaken
the theoretical analysis. Recent developments have further
improved accuracy in the experimental measurement of in
vivo knee kinematics. These developments allow significant improvements upon previous limitations by applying
patient-specific task-dependent models in the study of joint
pathogenesis.
The vast majority of dynamic knee studies have been
performed with conventional motion analysis techniques,
using markers attached to the skin. Conventional motion
analysis is not sufficiently accurate to enable analysis of
cartilage stress. Previous studies have shown that skin
markers move substantially relative to underlying bone,
with RMS errors of 27 mm and peak errors as large as
14 mm in estimates of tibial position during gait (12). A
study of four subjects during running and hopping (using
250 frames1 stereo radiography) has demonstrated skin
marker motion relative to the femur averaging 15 mm
throughout the motion, with peak-to-peak errors of the
oscillation at impact averaging 714 mm (13). Errors were
both subject and activity specific (14). Techniques have
been developed for improving estimates of bone dynamics
from skin markers using large numbers of markers and
optimizationmodeling of soft tissue deformation (15,16)
but the performance of these methods for in vivo studies
has only been validated for the tibia (during a slow, impactfree 10 cm step-up movement of a single patient with an
external fixator). Average errors were low, but peak errors
routinely exceeded 1 mm. Errors would likely increase
significantly during faster movements and movements
involving impact, and also where the soft tissue layer
between skin and bone is thicker (e.g., for the femur).
However, if the kinematic measurements are to be used
in conjunction with musculoskeletal models to estimate
dynamic loads and stresses on joint tissues, then even
errors as small as 1 mm may be unacceptable. For example,
when estimating strains in the ACL, a 1 mm error in
tibiofemoral displacement could introduce uncertainty in
the ligament length of approximately 3% (assuming a
nominal ligament length of 30 mm). This error is similar in
magnitude to estimated peak ligament elongation occurring during common activities, such as stair climbing (17).
For investigating cartilage deformation, this error magnitude would be even less acceptable. A 1 mm displacement
213
would be equivalent to a cartilage strain of 25%, relative

to the average thickness of healthy tibiofemoral cartilage
(18). A displacement error of this magnitude would translate into huge differences in estimates of contact forces.
Thus, efforts to model, predict and correct for soft tissue
deformation are unlikely to achieve sufficient accuracy for
assessing soft tissue behavior. Alternatively, kinematics
from a high speed stereoradiographic system capable of
tracking implanted tantalum markers in vivo with 3D
accuracy and precision better than 0.072 mm in translation
and 0.358 in rotation (19) are more appropriate in use with
advanced computational models. This accuracy is an order
of magnitude or greater of improvement over conventional
motion analysis techniques, and is uniquely capable of
providing the accuracy necessary to model joint stresses.
From Experimental to Advanced Theoretical Analysis in Joint
Mechanics
In addition to measuring joint kinematics and contact
areas, investigators have attempted to measure articular
contact stresses and pressures. However, stresses throughout the cartilage layer cannot be measured experimentally.
Direct measurements of stress can be made at the articular
surface using pressure sensing devices (2022) (e.g., pressure sensitive Fuji film, piezoresistive contact pressure
transducers, dye staining, silicone rubber casting). For
cadaver studies, Fuji film sheets (Fuji Prescale Film; Itoh,
New York, NY) are inserted in a joint and if pressed
produce a stain whose intensity depends on the static
applied pressure. Alternatively, digital electronic pressure
sensors (e.g., K-scan, Tekscan, Boston, MA) can be placed
onto the articular surface. These sensors are thin and
flexible, and can be made to conform to the anatomy of
the medial and lateral knee compartments. They consist of
printed circuits divided into grids of load-sensing regions.
Each load-sensing region within the grid has a piezoresistive pigment that can be used to determine the total
compressive load within that region. After appropriate
calibration procedures, dynamic pressure distributions
can be calculated. In addition to providing a continuous,
dynamic readout, K-scan has been reported to more accurately estimate contact areas than Fuji film (23,24).
There are significant concerns with the use of these
sensors for estimating actual contact pressures. These
techniques measure only surface-layer stresses, they alter
the nature of cartilage surface interactions and are too
invasive for in vivo human use. Thus, the clinical validity of
articular pressure measurement with such sensors is questionable. They can, however, be important tools for the
evaluation of the predictive power of joint models (2). By
including the sensor in a finite element model, the effects of
the sensor film on the actual contact mechanics can be
accounted for (25). Thus, contact pressure predictions from
such models can be directly compared to the pressure sensor
measurements for finite element (FE) model validation.
Many in situ experimental studies have been conducted
to obtain 3D knee joint kinematics and force-displacement
data (21,2630). Cadaver studies, however, cannot reproduce the complex loading seen by the joint during strenuous movements, since the muscle forces driving the
214
movement cannot be simulated. Because of these fundamental limitations of experimental measures, mathematical models are favored for obtaining comprehensive
descriptions of the spatial and temporal variations of cartilage stresses. A numerical model could be used to perform
parametric studies of geometry, loading or material properties in controlled ways that would not be possible with
tissue samples.
Theoretical Analysis of Joint Mechanics
During the last two decades, a number of theoretical joint
mechanics studies with different degrees of accuracy and
predictive power have been presented in the literature (31
39). Computational modeling work has included anatomical or geometrical observation (40,41) and analytical mathematical modeling (4246). More recently, advanced FE
modeling approaches allowed for improvements in the
predictive power of localized tissue deformation (4754).
Joint biomechanics problems are characterized by moving
contacts between two topologically complex soft tissue
layers separated by a thin layer of non-Newtonian synovial
fluid. A prime example is the multibody sliding contact
problem between the tibia, femur, and menisci. The complexity of such problems requires implementation of
sophisticated numerical methods for solutions (5558).
The finite element method is ideally suited for obtaining
solutions to joint contact problems. Thus far, much of the
finite element analysis has been applied to the study of
hard tissue structures, often as it relates to prosthetic
devices (59,60). When addressed, soft tissue layers are
treated as single-phase elastic materials. As a consequence
of the relative dearth of precise patient specific geometric
data, material properties and insufficiency in accuracy of in
vivo kinematics for input, no patient specific computational
models have been reported for longitudinal clinical joint
studies.
Surface Modeling
Surface modeling methods calculate the shape variations
of joints and visualize the proximity of subchondral bone
surfaces during static loading or dynamic movement.
These methods can combine in situ data, motion analysis
optical system data or high speed biplane radiographic
image data and 3D bone surface information derived from
computed tomography to determine subchondral bone
motion. This method can be used to identify the regions
of contact during static loading or dynamic motion, to
calculate the surface area of subchondral bone within
close contact, and to determine the changing position of
the close contact area during dynamic activities (Fig. 9).
In vivo dynamic joint surface interaction information
would be useful in the study of osteoarthritis changes in
joint space and contact patterns over time, in biomechanical modeling to assist in finite element modeling, and in
identifying normal and pathological joint mechanics preand postsurgery. Previous attempts to quantify the interaction between bones have utilized various methods
including castings (62,63), pressure sensitive film (64),
mathematical surface modeling (65,66), implant registra-
Figure 9. Example applications showing dynamic in vivo tibiofemoral bone surface motion using joint proximity (Euclidian
distance) mapping during human one-legged hopping.
tion (67) and cine phase contrast magnetic resonance

imaging (MRI) (68). The casting method can only be
applied to cadaver models under static loading conditions.
Pressure sensitive film also requires a cadaver model and
necessitates inserting material into the joint space. Mathematical surface modeling allows analysis of dynamic
motions in vivo, however, the joint must be disarticulated
after testing. Implant registration requires either surgical
implants or nonsubject specific image matching algorithms. Cine phase contrast MRI requires repeatedly performing the same motion pattern during testing and is
limited to a small range of motion. The process described
below is an improvement on these previous techniques
because it utilizes live subjects performing dynamic tasks
with unrestricted motion. Direct measurement of articular
cartilage behavior in vivo during dynamic loading is problematic. In order to estimate the behavior of articular
cartilage, the surface proximity interaction method that
precisely tracks the motion of subchondral bone surfaces in
vivo. Articular cartilage behavior is then estimated from
these subchondral bone measurements.
Anderst et al. (61) described a method to estimate in vivo
dynamic articular surface interaction by combining joint
kinematics from high-speed biplane radiography with 3D
bone shape information derived from computed tomography (CT). Markers implanted in the bones were visible in
both the CT scans and the radiographic images, and were
used to register the subchondral bone surfaces with the 3D
bone motion. Joint surface interactions were then estimated by analyzing the relative proximity of the subchondral bone surfaces during the rendered movement.
Computed Tomography data can be also used for joint
geometryshape characterization. The method is referred
as reconstruction of volumetric models into rendered joint
surface geometry models.
Computed Tomography data are typically collected for
this method with slice spacing between the different
images of 0.625 1.25 mm and the in-plane resolution
215
The Joint Distribution Problem
Figure 10. Articular surface matching (femoral condyle on top

and tibial plateau below) using geometrical objects (69).
is 0.2930.6 mm depending on the size of the bone. The

CT scans are reconstructed into 3D solid figures using
software that employs reconstruction techniques, that is,
the regularized marching tetrahedra algorithm by Treece
et al. (70). If necessary, threshold values are adjusted to
ensure the entire bone surface appeared in the reconstruction and the opposing bone surfaces never overlapped in
computer animations of the motion.
Anteriorposterior and lateral radiographs are commonly used to preoperatively determine prosthetic size
and proper donor selection for osteochondral allografts.
By using 3D computer aided design tools and the reconstructed 3D joint geometry from CT described above, size
determination is less prone to out-of-plane imaging errors
associated with sagittal and coronal roentgenograms.
Assessment of surface size, curvature analysis and knee
incongruity is possible with in vivo CT [Fig. 10 (69,71,72)].
After the 3D joint surface reconstruction models the
distal articular femur (n 16) can be represented by six
circles, the diameters of these circles, their angular arcs,
and the distances between their centers varied with the
size of the femur (Fig. 11; Table 5). There is a statistically
significant association between several geometry parameters when the lateral or the medial distal femur is
studied independently. These associations do not exist
when we correlate medial versus lateral compartments
across the population.
DP
DA
(b)
DC
X
Y
A
F
Anterior
Posterior
Figure 11. All the sagittal view measured parameters in the

study of femoral head congruity (69).
Much attention has been devoted to the solution to what

has become known as the general joint distribution problem that is, the problem of estimating the in vivo forces
transmitted by the individual anatomical structures in the
joint neighborhood during some activity of interest (73).
The prediction of forces in joint structures has many
applications. In the field of medicine, these predictions
are useful for obtaining a better understanding of muscle
and ligament function, mechanical environment within
which prosthetic components must operate, and mechanical effects of musculoskeletal diseases. In the realm of
sport biomechanics, these predictions are useful for better
understanding of the kinetic demands, performance
constraints, and mechanisms for improving athletic
performance. Industrial applications include the optimization of occupational performance and safety considerations. Although the general techniques for predicting
forces in joint structures may be used throughout this
broad range of applications, the particular method of choice
and the details of the analysis depend on the application.
The general force distribution problem normally arises
in the following way. The musculoskeletal system or a
relevant portion thereof is modeled as a mechanical system
consisting of a number of essentially rigid elements (body
segments) subjected to forces due to the presence of a
gravitational field, and segmental contact with external
objects, neighboring segments, and soft tissue structures
that produce and constrain system motion. The associated
inverse dynamics problem is then formulated and solved
to determine the variable intersegmental (joint) force
and moment resultants during the activity of interest.
The joint resultants are abstract kinetic quantities that
represent the net effect of all the forces transmitted by
the anatomic structures crossing the joint. At a typical
joint, these forces normally include the forces transmitted
by the muscles, ligaments, and articular (bony) contact
surfaces.
The unknown forces transmitted by the joint structure
are next related to the known intersegmental resultants by
writing the joint equilibrium equations. These equations
express the fact that the vector sum of all the forces in the
individual anatomic structures, and the vector sum of all
the moments about the joint center produced by those
forces, are equal to the intersegmental resultant force
and moment, respectively. Assuming that all joint geometry (point of application and orientation of forces) is known
and that these two independent vector equations (or six
independent scalar equations) involve as unknowns the M
muscle and L ligament forces, together with the 3C scalar
components of the C bony contact forces, these joint equilibrium equations are indeterminate whenever the sum
(M L 3C) of the unknown forces exceeds six. Thus, if
the system model includes only one bony contact force
(C 1) and more than three muscle and/or ligament forces
(M L > 3), the corresponding joint distribution problem
will be indeterminate and therefore have an infinite number of solutions.
Finally, the joint resultants are decomposed or distributed to the individual joint structures at each instant of
216
Table 5. Femoral (n 16) Medial and Lateral Compartment Measurementsa
DC mm
DA mm
DP mm
XY mm
F8
E8
Medial
SD
Lateral
SD
Ratio M/L
68.28
42.45
40.364
24.519
100.374
151.168
5.003
10.086
1.231
2.686
10.572
10.9
67.839
44.41
41.212
23.529
102.631
139.629
5.865
4.608
3.069
3.069
5.834
12.509
1.006
0.955
0.979
1.042
0.978
1.0824
interest during the activity, using some appropriate solution methodology.

The general joint distribution problem may thus be
stated in the following way. At any instant of time when
the joint resultants are known, the forces transmitted
by the individual joint structures are determined such that
the equilibrium equations, and all relevant constraints on
the forces in the individual joint structures, are simultaneously satisfied. The classical studies of joint distribution
problems use essentially two different methods to solve the
indeterminate joint distribution problem: the reduction
method, and the optimization method.
The mathematical modeling of human anatomy and its
functions has been influenced by two main simulation
approaches or philosophies. In the first the joint structures
are of no importance in the mathematical modeling while
in the second simulation of the geometry and structural
relationships of the joint components in addition to their
behavioral properties are the main tasks. Hefzy et al.
categorized these different approaches as phenomenological and anatomical, respectively (74).
Phenomenological Joint Models. The phenomenological
models include two groups: the rheological models and the
advanced figure animation models. The rheological models
analyze the dynamic behavior of a system by treating it as
viscoelastic, being composed of springs and dampers. However, the noncorrespondence of these components to the
structure of the components in the knee leads to no structural information in the model output.
The advanced figure animation models provide information on body dynamics by taking into account body segment
dimensions, masses, moments of inertia, and so on, but do
not model the detailed geometry of joints.
Anatomical Models. Two different approaches to modeling categorization exist in the experimental literature.
According to the first, the categorization of the models
depends on the type of motion reproduced by the mathematics. The second approach categorizes models according
to their structural basis. There is a vast number of studies
attempting to model specific component structure and
behavior that will be evaluated in order to identify the
optimum method for the modeling of each specific component.
The Reduction Method. The reduction method reduces
the number of unknown forces to correspond with the
number of equations governing the distribution problem
(or increases the number of equations to agree with the
number of unknowns, e.g., the deformation-force relations

for the unknown forces). For the general 3D distribution
problem, this implies that the number of unknown scalar
forces (M L 3C) must be reduced to six to allow for a
unique solution. Previous investigators have reduced the
number of unknowns by (1) grouping muscles and ligaments with apparently similar functional roles, (2) grouping multiply connected bony contact force regions, (3)
assuming a direction for the unknown bony contact force,
(4) using EMG data to determine when a muscle is active,
(5) ignoring ligament forces except near the limits of the
range of motion, and (6) ignoring antagonistic muscular
activity. Several models [(73,75); Fig. 12a, b] predicted
muscle forces and the bony contact force at the hip during
locomotion. In these studies, the indeterminate distribution problem at the hip or the knee was made determinate
through several simplifying assumptions. The hip muscles
were combined into six functional groups (long flexors,
short flexors, long extensors, short extensors, abductors,
and adductors), and ligament function at the hip was
assumed to be negligible. The forces transmitted by these
six muscle groups, combined with the three components of
bony contact force, comprised nine unknown scalar quantities. Only two of the three components of the resultant hip
moment were considered; analysis of the component tending to internally or externally rotate the femur relative to
the pelvis was rejected as inaccurate. Previous reports of
EMG activity were to demonstrate that there is little
antagonistic muscle action, and only muscle agonists were
considered. Despite these simplifications, the possibility of
activity in both the long and short flexors and extensors
still made the problem indeterminate. A solution was
obtained, however, by assuming activity in only one flexor
(either long or short, but not both) and one extensor.
The Optimization Method. The previous discussion
indicates that the distribution problem at a joint is typically an indeterminate problem, since the number of muscles, ligaments and bony contact regions available to
transmit forces across a joint in many cases exceeds the
minimum number required to generate a determinate
solution to the joint equilibrium equations. Determinate
solutions are obtainable only with significant simplification of joint function or anatomy. In contrast, the optimization method of solving the general joint distribution
problem does not require such simplification. Rather, it
retains many of the anatomical complexities incorporated
in defining the problem, and seeks an optimum solution
(i.e., a solution that maximizes some process or action).
Optimization techniques may be divided into linear and
217
Figure 12. Forward and inverse analysis driven mathematical models, (a) the Strathclyde model
animated using SIMM Musculographics-Motion Analysis Software. (From Ref. (77)) (b) Force
distribution method-inverse dynamics, the vector of bony contact force is shown on the tibial
plateau. (From Refs. (71) and (77)). (c) the Delft forward and Inverse shoulder model (78).
nonlinear methods Simultaneous use of linear cost and

constraint functions in the model formulation constitutes
the so-called linear optimization method. In contrast, if
the cost function and/or one or more of the constraint
functions is nonlinear, solution of the problem requires
the use of nonlinear optimization methods. Both of these
optimization methods have been made practical with
high speed computers, since the techniques require many
iterations of equation-solving to find an optimum solution.
Neuromuscular function is complex and poorly understood at the conceptual, if not detailed level. However,
it is generally assumed that physiological functions are
optimized in some way. In 1836, the Weber brothers
commented that we walk and run in the way that affords
us the least energy expenditure for the longest time and
with the best results. Experimental evidence suggests
that oxygen consumption (and presumably energy expenditure) is minimal at freely selected walking speeds,
supporting the assumptions of optimal physiological
neuromuscular function.
The optimization criteria that the neuromuscular control system chooses, either consciously or unconsciously,
to select muscle action may vary considerably with the
nature of the physical activity to be performed and the
physical capabilities of the individual. For example, muscle
control in sprint running may serve to maximize velocity,
while in walking the control process may serve to maximize
endurance. In a painful pathological situation, such as
degenerative joint disease, muscular control may serve
to minimize pain. If this pain occurs due to the joint surface
pressure, the appropriate optimization criterion may be to
minimize to bony contact force. Muscular control may also
serve to minimize the forces transmitted by passive joint
structures such as ligaments. These examples indicate that
there are possible optimization criteria to choose from, and
the choice of a criterion to solve a particular distribution
problem may not be obvious.
Given that gait presents a relatively unambiguous performance criterion, one can claim that it fits into the
framework of optimum control theory. Additionally, gait
presents a characteristic bilateral symmetry that leads to
relatively simple representation of the dynamic system.
Activity in the stance phase of gait is described by equality

constraints on the states, knowing that each gait stage
involves dynamic constraints that reflect the particular
nature of the phasic activity. However, our motivation in
the use of optimum control for the study of movement relies
upon the belief that it is currently the most sophisticated
methodology available for solving extremely complex problems. Optimal control theory requires not only that the
system dynamics are precisely determined and formulated
but also that an appropriate performance criterion is chosen. Therefore, deficiencies in the modeling that are present in either the system dynamics or the performance
criterion are indicated by differences between model and
experiment.
Forward Analysis. The technique of forward simulation
allows the study of the causal relationship between
forces acting on a mechanical structure and the resulting
movement (79). A first approach requires the description
of joint moments, initial positions and velocities for each
body segment as input variables. The forward dynamics
problem then is expressed as a set of differential equations
of motion with associated restraints and solved to yield
angular accelerations. Angular velocities and displacements are then determined by integration (80). Modifications to input variables, either joint moment profiles
or initial segmental configuration, can then be introduced and the resulting changes in the movement pattern
evaluated.
Due to recent improvements in musculoskeletal modeling, forward simulation driven by muscle activity rather
than joint moments is now possible. The definition of
muscle activation sequences and the initial configuration
of the mechanical system (angles and angular velocities)
are entered into the forward simulation. A physiological
model describing muscle excitation characteristics as
well as muscle mechanics is used for calculating the individual muscle force production and the resulting motion is
calculated.
Because of the inherent complexity of the musculoskeletal model and the multiple interaction of parameters,
forward simulation alone is unlikely to contribute to the
218
identification of aberrant muscle action which affects timing or force production, causing an aberrant gait pattern.
Matching the system response to the joint movement
profiles observed in an individual patient would require
extensive trial and error with no guarantee of ever finding
the parameter setting responsible for the aberrant joint
movement.
Optimization has been used previously as a curve-fitting
tool in addition to forward simulation techniques to reproduce the movement pattern observed in an individual
subject. In the studies of Chao and Rim (81), and also
Townsend (82), optimization techniques were used to
determine the applied joint moments by iteratively varying
them until the theoretical limb displacement fits those
measured in the laboratory. Chou et al. (83) predicted
the minimum energy consumption trajectory of the swing
limb using a similar approach. Using approximate muscleforce and/or joint moment trajectories, Pandy and Berme
(84,85) evaluated body segment motion and ground reaction forces during single stance, based on a 3D model
incorporating 7 DOF. In Jonkers et al. (76), initial inputs
to the forward simulation process were the normalized
quantified muscle activation patterns of 22 muscles, and
the initial segmental configuration (both angles and angular velocity) derived from Winter (86). Two distinct musculoskeletal models (one including 6 DOF, the other 7
DOF) were defined and a muscle driven forward simulation
was implemented. A series of optimization sequences then
were executed to modify the muscle activation patterns and
initial segmental configuration, until the system output of
the forward simulation approximated the angle data
reported by Winter (86). The accuracy and effectiveness
of the analysis sequence proposed and the model response
obtained using two distinct musculoskeletal models were
verified and analyzed with respect to the kinesiology of
normal walking.
Based on the integrated use of optimization and forward simulation techniques, a causal relation between
muscle action, initial segmental configuration and resulting joint kinematics during single limb stance phase of
gait was successfully established for a musculoskeletal
model incorporating 22 muscles and 7 DOF (Fig. 12).
Despite the inherent simplifications of the planar models
used in this study, several kinesiological principles of
normal walking were confirmed by the analysis and a
reference base for exploring the causal relation between
muscle function and resulting movement pattern was
established.
Finite Element Analysis of Human Joints. Adaptation of
finite element analysis (FEA) to the analysis of stress
in human joints, requires fundamental studies in the
following areas: (1) three dimensional FEA of moving
contact problems utilizing finite deformation laws (linear
elastic versus biphasic) for cartilage and non-Newtonian
laws for synovial fluid; (2) automated adaptive methods
for the generation and control of 3D computational models
using error estimates and controls that account for
nonlinearities, singularities, and boundary layer effects;
(3) improvements in geometry of FE models by using
patient specific MRI and CT imaging data. Recently
available semiautomatic 3D mesh generation tools

(reconstructed mesh from the volumetric imaging data),
and numerical methods for processing the material and
geometric data required for the contact analysis considerably reduce the time requirements of such efforts;
(4) implementation of recently available high accuracy
3D joint in vivo kinematics to calibrate the FE models,
and to assist in simulating strenuous activities; (5)
Parallel solution algorithms for the nonlinear timedependent problems utilizing high performance computer
architectures.
Several models of articular cartilage have been proposed to describe its mechanical behavior; however, none
of these models have been able to address the full spectrum
of this tissues complex mechanical responses. Typically,
these models implement a subset of known tissue behavior,
for which material properties must be determined from
experiments. Experimental results indicate that cartilage
exhibits flow-dependent viscoelasticity, anisotropy, and
tension compression nonlinearity due to its ultrastructure
and composition (87). These characteristics are further
compounded by the inhomogeneity of the tissue through
its thickness. It is widely accepted that the time-dependent
response of cartilage can be accurately represented by the
biphasic theory derived by Mow et al. (21). Under isotonic
conditions, this biphasic theory of incompressible solid and
fluid phases is appropriate for most applications involving
cartilage modeling for infinitesimal or finite deformations
(26,88). Numerical methods are required to solve these
nonlinear problems, even for relatively simple geometries.
Linear 3D elements (89) and nonlinear 3D formulations
have been presented for the biphasic theory (9092). However, full nonlinear 3D analysis for joints of realistic geometry and strains remains a computationally challenging
problem.
Hirsch (1944) proposed to use the Hertz contact theory
for contacting elastic spheres to model cartilage indentation (28). Askew and Mow (1978) analyzed the problem of a
stationary parabolically-distributed normal surface traction acting on a layered transversely isotropic elastic medium to assess the function of the stiff surface layer of
cartilage (29). More recently, complex, biphasic models
have been developed that target slow, quasistatic loading
response (89,90,9395). Under high loading rates, however, the cartilage demonstrates a mechanical behavior
that does not deviate substantially from a linear elastic
model (38,96), so complex and numerically unwieldy biphasic models may not be necessary for rapid loading events.
Eberhardt et al. (1990) (36) and (1991) (97) developed a
solution for the contact problem of normal and tangential
loading of elastic spheres, with either one or two isotropic
elastic layers to model cartilage. A modified Hertzian (MH)
theory that takes into account the thickness effect of the
elastic layer has been used for articular joint contact
analysis by several investigators (23,32,37,97). The results
from the studies listed above have demonstrated that
considerable differences may be found in cartilage stress
predictions depending on the particular cartilage constitutive model being employed, that is linear elasticity theory versus linear biphasic theory. In view of the above, it
has been proposed that it is adequate to use a transversely
219
Figure 13. (a) Knee FE model components shown in different color-implicit

formulation (102) (b) the knee FE model
during simulation of single leg-hoppingexplicit formulation (103).
isotropic, linearly elastic, homogeneous constitutive relationship to model the normal contact pressure distribution
of the tibial plateau (33,98).
In the case of FE modeling applied to impacts and very
rapid movements, model formulation problems must be
accounted for by the solution. Implicit techniques are not
widely used as they can be limited by problems of convergence. This method does not use kinematics to verify
whether contact occurs at the calculated points at the
associated load levels. For those applications, explicit
techniques have proven useful in their ability to simulate
deformations and contact conditions simultaneously
as well as large segmental displacements (70,99101)
(Figs. 13b and 14). Explicit finite element analysis was
also proven to be a valuable tool when simulating total
knee replacement motions due to loads applied by a knee
simulator (44,45). Stability of solution and low computational cost are the two main advantages of the explicit
methods over classic implicit techniques when forces are
used to drive the models (44,98,103109). The main limitation of the explicit technique is that the method is only
conditionally stable. Combination of both techniques is
suggested so that the explicit formulation will be used to
educate the implicit algorithms. Implicit formulations

are used when the localized effect of articular stress
needs to be addressed at the cartilage level with increased
subject specific refined geometry since these techniques
are more appropriate for the task (Fig. 13a).
Toward Patient-Specific and Task-Dependent Morphological FE Models. In order to apply FE computational methods to problems of joint mechanics, precisely measured
anatomical data must be used to construct a 3D solid model
from which the appropriate finite mesh can be constructed
and analysis performed (110,111). Computed tomography
(112) and MRI (113,114), or reshaped digital calipers and
machinecontrolled contact digitization (23) are commonly
used methods to obtain softhard tissue geometry. Clinical
modalities of these imaging techniques have resolutions that
are typically limited to 500 mm. Considering that the articular cartilage of the knee joint is only 4 mm thick (14), a
resolution of 500 mm is generally inadequate and special
protocols are required for the imaging procedure (19). Note
that all properly calibrated 3D FE knee joint models to date,
have been developed from images of cadaveric knees (43,115).
Models with patient specific geometry are possible using
Figure 14. Tibial plateau pressure at 10

and 30 degrees flexion during single-legged
hopping (103).
220
combined appropriate MR and CT imaging modalities that

will result in increased model detail (100).
With recent technological advances in 3D imaging,
computer hardware, and FE formulations, it has become
possible to estimate human tissue properties (100,103).
Estimation of the stiffness of hard tissue (i.e., cancellous
bone) using large-scale finite element (LSFE) models or
microFEs at both the apparent and trabecular levels is
therefore possible (47,52,70,116122). This direct analysis
approach has advantages over traditional statistical methods in that (1) all morphometric measurements available
from the 3D image data are contained in the FE model, and
(2) the FE model of the structure is a mechanistic rather
than statistical approach to the prediction of biological
tissue properties. Application to these techniques in the
clinical environment with in vivo data is gaining considerable ground. These recent improvements in imaging, allow
the generation of 3D models in a semi-automatic fashion
(23,69,123,124). Models now provide the opportunity to
study the mechanical effects of individual geometrical
variation (125). Such subject specific finite element
models are mainly useful under two conditions: (1) In
the statistical analysis, if the differences between subjects
are large relative to the differences in model predictions
related to the choice of input parameters that can only be
estimated. (2) In a contact FE problem with high accuracy
3D kinematics refined meshes are needed in the vicinity of
the contact zone (114).
The inherent ability of FEM to overcome geometrical
and topological modeling problems is no longer burdened
by some of the limitations apparent in early studies. Such
limitations have been overcome by recent computational
and experimental advances. These limitations included
insufficient accuracy of in vivo kinematicskinetics input
data, inadequate hardsoft tissue imaging, excessive computational cost and the inherently complex process of
material characterization in relation to some of the components of the stress tensor in biological material.
JOINT STABILITY
Newtons third law states that forces always exist in pairs
that are equal and opposite in direction, such that if one
body pushes against another, the second body will push
back against the first with a force of equal magnitude if the
state of motion is constant. This would mean that in order
for muscles to be able to pass movement on to a limb
segment, there has to be an interaction between the segment and another bone, and a joint structure that will
allow the desired rotational direction and force. Such a
mechanism prevents translational movements, which can
dislocate one rigid bony limb segment from another.
Bones, ligaments, and muscles all contribute to joint
stability. The extent each structure participates in maintaining joint stability differs between joints. Because bones
are rigid relative to the other tissues in the joint, they can
provide great stability to the joint. In general, the greater
the circumference of the bony segment enclosed by its
counterpart, the greater the amount of translational stability that exists in the joint. For example, the femoral head
is almost completely enclosed by the acetabulum in the hip
joint, whereas the humeral head is only slightly enclosed by

the glenoid in the shoulder. It is obviously easier to dislocate the shoulder than the hip.
Ligaments, much like cables, can restrict rotational or
translational motions depending on their location and the
direction of the force. For example, the cruciate ligaments
limit the anterioposterior translation of the tibia on the
femur at the knee joint, while talocalcaneofibular ligaments prevent rotational motions as well as translational
motions between talus, fibula and calcaneus. However,
ligaments, unlike bones, cannot provide rigid constraints
as they are relatively soft and flexible.
Musclestendons are also soft and flexible and, like
ligaments, provide nonrigid constraints to the joint. A
significant difference between the two, however, is that
muscles are active in controlling the joint motion whereas
ligaments are only passive stabilizers. Muscle action is
usually complementary to that of ligaments and, in fact,
can protect ligaments from damage or can protect the joint
from further damage in case that the ligament is ruptured.
Tensile forces exerted on muscles are counterbalanced by
compressive forces across the joint surfaces providing stability to the joint against forces acting to open up the joint
space. Thus, muscles can support or limit motions and
serve the dual function of providing desired movements
while contributing to joint stability.
The Hip Joint
The hip joint is the link of the upper body and the pelvis
trunk with the lower limbs, the main locomotion facility of
the body. It is a ball-and-socket joint (Table 4) in which the
head of the femur resides in the acetabulum of the pelvis,
making one of the largest and most stable joints in the
body. The surface area and the radius of curvature of the
articular surface of the acetabulum closely match that of
the articular surface of the femoral head. The hip joint is
structurally a highly constrained joint. Because of the
inherent stability conferred by its bony architecture, this
joint is well suited for performing the weightbearing supportive tasks that are imposed on it.
The femoral ball is embraced by the acetabular socket,
allowing rotation to occur with virtually no translation.
The cartilage that covers the acetabulum thickens peripherally (126). A plane through the circumference of the
acetabulum immediately at its opening would project with
the sagittal plane intersections at an angle of 408 (opening
posterior) and 608 (opening laterally). This architectural
constraint imparted by the bony shapes almost eliminates
the need for ligamentous and soft-tissue constraints to
maintain the stability of the hip articulation. Although
this increased constraint provides stability to the hip,
there is a structural drawback. Such constraint limits
the global range of motion of this joint at the fulcrum of
the lower extremity. Fortunately, human biomechanics
and everyday tasks performed by the hip do not violate
these limitations and the hips range of motion is very
rarely subjected to extremes. During most ambulatory
activities, such as normal bipedal locomotion, the lower
extremity is positioned anteriorly in the sagittal plane
with only small rotations necessary in the other two
planes. Hip flexion of at least 1208, abduction of at least

208, and external rotation of at least 208 are necessary for
carrying out normal daily activities. Activities such as
descending stairs sitting, rising from a chair, and dressing
require greater degrees of flexion and rotation at the hip
joint. For example descending stairs requires 368 of
motion whereas squatting requires 1228 of motion in
the sagittal plane (127).
The hip joint reaction force (at a neutral position) can
reach three times body weight (BW) in single legged
stance, but could get up to six times BW during the stance
phase of gait and increases significantly with gait velocity. The main mechanism influencing this magnitude is
the ratio of the abductor muscle force and the effect of
the gravitational force moment arms. The rule normally
suggests greater reaction forces expected at low ratios
(128). Bracing and use of a cane can decrease the hip joint
force.
The Knee Joint
The knee consists of a two joint structure: the femorotibial
joint and the patellofemoral joint. The femorotibial joint
is the largest joint in the body and is considered to be a
modified hinged joint containing the articulating ends of
the femur and tibia. The patellofemoral joint consists
of the patella, the largest sesamoid bone, and the trochlea
of the femur. Taken together, the knee joints function to
control the distance between the pelvis and the foot as a
control link. Because of the role of the knee in weightbearing, its surrounding of very strong musculature and
its location between the two longest bones in the body,
tremendous forces are generated across it. Surface motion
occurs simultaneously in both sagittal and the transverse
plane with the first being the dominant plane of motion
(129). During activities such as running, landing, and
pivoting, the knee functions to maintain a given leg
length and acts as a shock absorber. During stair climbing, crouching, and jumping, large propulsive forces at
the knee (several times BW) are generated to control the
degree and speed of shortening and lengthening of the
leg. In these situations, knee stability is a dynamic process maintained through fixed bony and ligamentous
constraints and modified by the action of the muscles
crossing the joint. The higher the rate of the loading
the more demanding the role of the knee in sustaining
221
stability. The bony morphology of the femur, tibia,

menisci, cruciatecollateral ligaments, and patella contribute to joint stability, but to a lesser extent than that of
other more constrained joints such as the hip. Static
constraints play a very significant role in knee stability
as compared with the shoulder for example where stability is maintained through the dynamic action of the
surrounding muscles. The cruciate and collateral ligaments are the major structures limiting motion at the
knee. The posteromedial and posterolateral capsular complexes augment the four primary ligaments, with the
menisci playing a lesser role. The muscles crossing the
knee contribute to dynamic stability and are particularly
important in the presence of pathologic laxity. A method
that describes the constraining mechanism of the anterior
and posterior ligament has used the notion of the 2D
four bar linkage. The instant center of rotation, designated primarily by the femoral condyle surface shape,
follows a semicircular pathway, and the direction of displacement of the femorotibial contact points is tangential
to the surface of the tibia, indicating gliding throughout
the range of motion. However, the axis of rotation at the
knee does not remain fixed during flexion. Indeed, as the
knee flexes the screw axis will sweep out a ruled surface in
space, known as the axode. This fluctuation in the screw
axis signifies that the knee is not truly a hinge joint, for
which the axode would degenerate to a fixed line in space.
Usually, the knee is approximated as a hinge joint, a
simplification that may be acceptable for flexion angles
between 45 and 908 where the moving screw axis remains
very close to the line passing through the centers of
curvature of the two posterior femoral condyles. The
motion at the articular surfaces is not one of pure rolling,
but a combination of rolling and sliding as indicated by the
screw axis that never lies near the articular surfaces of
the tibiofemoral joint.
The Foot Structure; the Ankle Joint
The ankle joint can be described as a saddle-shaped lower
end structure of a long bone (tibia and fibula) Fig. 15. Its
inferior transverse ligament encloses the superior aspect
of the body of the talus (the trochlea). It is the joint that
first receives the transient impact that travels through
the tibia in gait or other movement. It alternates in both
form and function to receive load as a shock absorbing
Figure 15. The Wayne State University Ankle joint and foot Finite element model (130).
222
mechanism and to propel significant leverage based

force during fast locomotion. The subtalar and ankle joint
act like a mitered hinge. The tibial surface forming
the superior dome of the ankle is concave sagittally, is
slightly convex from side to side, and is oriented 938
from the long axis of the tibia (it is higher on the lateral
than the medial side). The upper articulating surfaces of
the talus appear to match closely that of the cavity formed
by the tibiofibular mortise. The superior part of the body of
the talus is wedge-shaped. It is about one-fourth wider in
front than behind, with an average difference of 2.4 mm;
anteriorly a minimal difference of 1.3 mm and a maximal
difference of 6 mm. From front to back, the articular surface spans an arc of 1058. This surface contour, having a
smaller diameter medially than laterally, has been compared to a section or rostrum of a cone. The primary
motion of the ankle joint is dorsiflexionplantarflexion.
Its axis of rotation is obliquely oriented with respect to all
three anatomic planes with ankle dorsiflexion and tibial
internal rotation being associated with subtalar eversion
(pronation) whereas the ankle plantarflexion and tibial
external rotation are associated with subtalar inversion
(supination). The axis extends from anterior, superior,
and medial to inferior, posterior, and lateral as it is
passing through the inferior tips of the malleoli. It is at
angles of 938 with respect to the long axes of the tibia and
128 to the joint surface. However, rather than a true
single ICR, the ankle has been noted to have multiple
instant centers, all of which perturbate and fall very close
to a single point within the body of the talus. Through a
complete arc of ankle rotation, the center may be displaced anywhere from 4 to 7 mm. The oblique orientation
of the axis of rotation to the sagittal, coronal, and transverse planes, translation of the talus in the mortise can
occur in all three directions. The talus has been observed
in vitro to rotate easily in the ankle mortise implying
relative movement between the malleoli. Because the
trochlea is wider anteriorly than posteriorly, it has been
suggested that lateral play of the talus within its mortise
occurs only when the ankle is in plantarflexion. Subtalar
motion has been described as screw-like influencing the
flexibility of the transverse tarsal joint. Others suggest
that instability exists in dorsiflexion, while others yet
believe that with intact ligaments translation, occurs only
in the sagittal direction. These differences can be
explained by behavior of >100 ligaments and by the roles
played by the subtalar joint, the kinematic chain of the
hindfoot, and the muscles that traverse this area in
transmitting forces across this area during plantarflexion
and dorsiflexion. The talus is unique because this bone
has seven articulations that connect it to four other bones,
it lies between the foot and the leg, and contains no
muscular attachments. The stability of the talus and its
articulations, therefore, relies heavily on the ligamentous
attachments and musculotendinous complexes that traverse the talus and attach distally. Therefore the main
characteristic of ankle joint is its strong passive stability
attributed to a variety of factors. First is the bony stability
provided by contact of the trochlea with the tibial plafond.
Second are the medial and lateral cartilaginous slightly
concave surfaces that articulate with the two malleoli.
Third are the ligamentous connections between the tibia,

fibula, talus, and calcaneus. Ankle stability increases
during weight bearing and depends more on articular
surface congruency.
The tarsometatarsal joints are relatively mobile and
intrinsically stable joints that produce the arch-like configuration allowing wide range of motion at the first metatarsophalangeal joint with gliding during most of its range
and jamming artful extension. The medial longitudinal
arch functions as a beam and a truss.
The mode of footankle mobility and muscular control is
the most significant determinant of both limb stability and
body progression (131).
Standing barefooted we load our heels with two-thirds of
the load while the other third is loading the forefoot.
During walking and at the early phase of stance, the center
of pressure moves from the posterolateral heel rapidly
across the midfoot, a phenomenon coupled with the firing
of the anterior tibial musculature to slow foot plantarflexion and prevent foot slap. Then, at mid- and late stance the
posterior calf musculature fires, propelling the body over
the foot towards toe-off phase where the hallux bears the
most pressure.
At higher speeds/rates of mobility it is the intrinsic
mechanical properties that provide control rather than
the neuromuscular control system. The force at the ankle
joint can reach magnitudes up to six times body weight
during walking and thirteen times BW during running. The
heel fat pad is a very effective shock absorbing mechanism
and it has been shown that high heels, or narrow shoes,
narrow toebox, can lead to altered foot mechanics that
ultimately result in foot deformities and pain.
The Spine
The spine is composed of a series of vertebrae (7 cervical, 12
thoracic, 5 lumbar, the sacrum and coccyx) and intervening
soft tissues, such as intervertebral disks, ligaments, and
muscle attachments. It provides important functions
including support of the body structure and protection of
vital tissues such as the spinal cord, nerves and arteries.
Yet it is flexible and allows mobility to the torso. Several
studies have described the passive and active range of
motion of spinal segments differently (Table 4). The motion
of the spine is usually analyzed through consideration of
motion segments that are comprised of two adjacent vertebrae with the intervertebral disk and other intervening
soft tissues. These structures, also called functional spinal
units (FSU), move with 6 DOF, however, the motion is
quite complex due to six articulate faces between the two
bony segments and attachment of multiple ligaments and
muscles. Normally simultaneous translations and rotations of FSU are coupled in the analyses (134). Creep,
relaxation and hysteresis are the three prevailing viscoelastic characteristics of the intervertebral disks (1). At high
rates of loading the disks serve as a shock absorber with
compression strength being higher from upper cervical to
lower lumbar levels. Although this is true for most joints,
vibrational properties of spinal segments have attracted
particular attention as they are thought to be related to
injury and pain. The spinal cord exhibits some longitudinal
223
Figure 16. The Wayne State University full body impact model and the spine model (130,135).
elasticity but very poor axial translation. These translational forces are the ones primarily associated with
neurological injury. Apparently muscular support is of
vital importance is stability. If for example bilateral facet
resection exceeds 50% the significant increase in annulus
stresses may occur. Seatbelts can adequately protect front
seat occupants against frontal impact in an airbag
equipped vehicle (135).
Several models of the spine have been developed
(130,135). A nonlinear finite element model of lumbar spine
segment L3L5 is shown in Fig. 16. The effects of upperbody mass, nucleus injury, damping, and different vibration frequency loads were analyzed for the whole body
vibration (WBV) using this model. Anterior regions of
L3L5 segment show small vibration amplitudes, but posterior regions show large amplitudes. The posterior regions
of intervertebral discs of lumbar spine are easy to injury
during long-term WBV compared with anterior regions.
The vibration of the human spine is more dangerous to
facets, especially during WBV approximating a sympathetic vibration, which may lead to abnormal remodeling
and disorders of lumbar spine.
stability mechanisms. Axial rotation of the humerus is

conventionally described by the degrees of internal or
external rotation when the humeral axis is parallel to
the thorax, or perpendicular to the thorax (abducted
908). The primary anterior stabilizers of the shoulder when
the arm is abducted at 908 are the inferior glenohumeral
ligaments. Horizontal abduction and adduction, also
known as horizontal extension and flexion, respectively,
are commonly used to describe arm position when the axis
of the humerus is perpendicular to the thorax. The muscles
surrounding the structure produce a barrier effect by
applying compressive forces and by eccentric contraction
resulting is active stability. The level of elevation of the
normal arm is 1671688 and 1711758 for men and women,
respectively. Average extension or posterior elevation is
approximately 608. When the arm is adducted by the side of
the body, it is possible to achieve approximately 1808 of
rotation. With abduction of the arm to 908, however, the
total arc of rotation is reduced to 1208. The range of motion
of the shoulder decreases with normal aging. The maximum loads that are characteristic for the glenohumeral
joint can reach magnitudes of one and a half body weight
(132).
The Shoulder
The shoulder complex allows the arm to move with respect
to the thorax. The biomechanics of the shoulder involves
the study of four different articulations, namely, the acromioclavicular, sternoclavicular, glenohumeral joints, and
the scapulothoracic joints. The shoulder complex has the
greatest range of motion among joints in the body (Table 4).
Traditional descriptions of humeral motion are based on
the angle between the humerus and the thorax in the
sagittal plane (flexion and extension) and the coronal plane
(abduction), with one very common characteristic: the
glenohumeral articulation is inherently unstable due to
the shallow nature of the glenoid fossa, containing only
one-third of the diameter of the humeral head. The ligaments, capsular and muscular structures are the main
The Elbow
The elbow joint consists of the proximal end of ulna and the
capitulum of the humerus. Three articulations are present:
Humeroulnar, humeroradial, and proximal radioulnar
(133). It is a hinge-type joint whose functions include
positioning and stabilizing the hand, providing lever-type
support during lifting and weight-bearing during activities, such as a pommel horse exercise. The humerus-ulna
joint itself has only one degree of freedom, however, the
rotational motion of the radius over ulna provides another
degree of freedom to the elbow. Thus the elbow can be
considered to have two degrees of freedom. Basically its
changing center of rotation during flexion-extension makes
it more than a hinge joint: flexion-extension (01608)
224
and axial rotation, or pronationsupination (70808 and

80858, respectively). The functional range of motion for
most activities of daily living is 1008 for flexionextension
(301308) and pronationsupination (508) at the elbow
joint. The instantaneous axes of rotation for flexionextension
are localized at the center of the lateral projected curvature of
the trochlea and capitulum. The size of this region measures
only 23 mm.
The axis of forearm rotation passes through the capitulum and head of the radius, extending to the distal ulna.
The carrying angle, the angle between the long axis of the
humerus and the long axis of the ulna, averages 78 in men
and 138 in women.
Motion in the plane of varusvalgus rotation at the
elbow indicates joint instability and such motions are
restricted. The rotational forces are mainly resisted by
the anterior and posterior oblique fibers of the medial
collateral ligament; the former is stretched during flexion and extension, whereas the latter is stretched
only during flexion. In addition to the medial collateral
ligament, the shape of the articular surface and the
anterior capsule contribute to the resistance to valgus
stress while the articular congruity and the radial head
provide stability in flexion. The same structures resist
valgus stress in extension as well. It is the lateral collateral ligament, anconeus, and joint capsule that provide
stability under varus stress. Force generated at the elbow
can reach up to three times body weight in everyday
activities (136).
The Wrist
The wrist or carpus is a very complicated joint consisting of
multiple articulations. It provides a stable support for the
hand, allowing for the transmission of grip forces as well as
positioning of the hand and digits for fine movements.
Wrist position affects the ability of the fingers to operate
(flex-extend) and to effectively grasp. The main function of
the wrist is to fine-tune grasp by controlling the length
tension relationship in the extrinsic muscles to the hand
(137). The hand is the principal instrument of touch and its
combination of sensibility and motor function has classified
it as the ultimate supportive organ of information, accomplishment, and evolution. The self-selected wrist position
for maximal power grip has been shown to be 358 of
extension with 78 of ulnar deviation. In addition, full wrist
flexion deteriorates the efficiency of the finger flexors and
the grip strength to 25% of that available when the wrist is
in extension. Martial arts experts have exploited this
knowledge for a long time to disarm assailants. When
the wrist is positioned in flexion, the combination of a
passive extensor tenodesis effect, increasing resistance
to flexion, and decreasing excursion of the flexors results
in a weakened grip. The assailant, therefore, is unable to
maintain a grip on the weapon. The strength of the finger
flexors is more than twice that of extensors. The unique
feature of the metacarpophalangeal (MCP) joints is their
structural asymmetry, which assisted by the ligament
positions and the bony configuration of the metacarpal
heads results in effective stabilization and finely tuned
motion.
OVERVIEW
Human diarthroidal joints can support high levels of
mechanical load for several decades, yet degenerative joint
diseases occur in epidemic proportions every year. Understanding the role of mechanics in diseases, such as osteoarthritis, requires analysis of the behavior of both healthy
and pathological joints under a full range of physiological
loading conditions. Relationships between joint function,
meniscus injury, and osteoarthritis after trauma (e.g., ACL
injuryreconstruction) requires detailed knowledge of
internal tissue stresses. Such knowledge is necessary to
advance our understanding of cartilageligamentmeniscus failure mechanisms at the macroscopic level, as well as
mechanotransduction at the cellular level.
Experimental measurements alone are not sufficient to
delineate the detailed biomechanics of the human joint.
Recent publications have confirmed that mathematical
modeling can be an effective tool for the simulation and
analysis of complex biological structures (e.g., the human
knee). However, existing modeling strategies, based on
generic geometry and/or driven with simplistic kinematics
and loading assumptions, have been of limited value in
addressing clinical hypotheses. Research efforts in the
biomechanics community have focused on cadaveric studies for the characterization of tolerances on material,
geometry, and attachment parameters that will restore
contact pressure distribution to within a specified deviation from accepted normal values. Models driven directly
by in vivo kinematic data have not been possible due to the
relatively poor accuracy available from traditional motion
analysis methods, limiting the predictive power of FE
modeling techniques in the study of knee pathogenesis.
New advances in modeling problem formulations that take
advantage of patient-specific geometry and unique motion
analysis systems (cineradiography) for high accuracy 3D
knee kinematics can overcome limitations of previous
approaches. This should generate high quality estimates
of internal tissue stresses, providing new insight into the
role of tissue loading in the development and progression of
musculoskeletal diseases such as osteoarthrosis.
BIBLIOGRAPHY
1. Maroudas A. Physical chemistry of articular cartilage and the
intervertebral disc. In: Sokoloff L, ed. The Joints and Synovial Fluid. New York: Academic; 1980: 239291.
2. Ateshian GA, Lai WM, Zhu WB, Mow VC. An asymptotic
solution for the contact of two biphasic cartilage layers. J
Biomech 1994;27:13471360.
3. Mow VC, Ratcliffe A. Structure and function of articular
cartilage and meniscus. In: Mow VC, Hayes WC, eds. Basic
Orthopedic Biomechanics. 2nd ed. Philadelphia: LippincottRaven; 1997. p 113177.
4. Wittenburg J. Dynamics of Systems of Rigid Bodies. Stuttgart, Germany: B.G. Teubner; 1977.
5. Woltring HJ. Planar control in multi-camera calibration for
3-D gait studies. J Biomech 1980;13:3948.
6. Woltring HJ. Representation and calculation of 3-D joint
movement. Human Movement Sci 1991;10:603616.
7. Goldstein H. Classical Mechanics. Reading (MA): AddisonWesley; 1980.
8. Dempster WT. Space Requirements of the Seated Operator.
Wright Patterson Air Force Base, OH, 1955.
9. Drillis R, Contini R, Bluestein M. Body segment parameters;
A survey of measurement techniques. Artif Limbs 1964;25:
4466.
10. Full RJ, Koditschek DE. Templates and anchors: neuromechanical hypotheses of legged locomotion on land. J Exp Biol
1999;202:33253332.
11. Nigg BM, Herzog W. Biomechanics of the musculoskeletal
system. West Sussex (UK): Wiley; 1999.
12. Manal K, McClay Davis I, Galinat B, Stanhope S. The
accuracy of estimating proximal tibial translation during
natural cadence walking: bone vs. skin mounted targets. Clin
Biomech (Bristol, Avon) 2003;18:126131.
13. Tashman S, Anderst W. Skin motion artifacts at the knee
during impact movements. Seventh Annu Meet, Gait and
Clinical Movement Analysis Soc. Chattanooga (TN); 2002.
14. Tashman S, Anderst W. Skin motion artifacts at the knee
during impact movements. Gait Posture 2002;16:1112.
15. Alexander EJ, Andriacchi TP. Correcting for deformation in
skin-based marker systems. J Biomech 2001;34:355361.
16. Andriacchi TP, Toney MK. A point cluster method for in vivo
measurement of limb segment movement. Adv Bioeng 1994;
28:185186.
17. Fleming BC et al. The strain behavior of the anterior cruciate
ligament during stair climbing: An in vivo study. Arthroscopy
1999;15:185191.
18. Shepherd DE, Seedhom BB. Thickness of human articular
cartilage in joints of the lower limb. Ann Rheum Dis 1999;58:
2734.
19. Tashman S, Anderst W. In-vivo measurement of dynamic
joint motion using high speed biplane radiography and CT:
Application to canine ACL deficiency. J Biomech Eng
2003;125:238245.
20. Jurvelin JS, Arokoski JP, Hunziker EB, Helminen HJ.
Topographical variation of the elastic properties of articular cartilage in the canine knee. J Biomech 2000;33:
669675.
21. Mow VC, Kuei SC, Lai WM, Armstrong CG. Biphasic
creep and stress relaxation of articular cartilage in compression? Theory and experiments. J Biomech Eng 1980;102:
7384.
22. Mow VC, Lai WM, Holmes MH. Advanced theoretical and
experimental techniques in cartilage research. Biomechanics: Principles and Applications. Boston: Martinus Nijhoff
Publishers; 1982. p 4774.
23. Fregly BJ, Sawyer WG. Estimation of discretization errors
in contact pressure measurements. J Biomech 2003;36:
609613.
24. Harris ML, Morberg P, Bruce WJ, Walsh WR. An improved
method for measuring tibiofemoral contact areas in total
knee arthroplasty: A comparison of K-scan sensor and Fuji
film. J Biomech 1999;32:951958.
25. Wu JZ, Herzog W, Epstein M. Articular joint mechanics with
biphasic cartilage layers under dynamic loading. J Biomech
Eng 1998;120:7784.
26. Vermilyea ME, Spilker RL. Hybrid and mixed-penalty finiteelements for 3-D analysis of soft hydrated tissue. Int J Numer
Methods Eng 1993;36:42234243.
27. Prendergast PJ, van Driel WD, Kuiper JH. A comparison of
finite element codes for the solution of biphasic poroelastic
problems. Proc Inst Mech Eng [H] 1996;210:131136.
28. Hirsch C. A contribution to the pathogenesis of chondromalacia of the patella. Acta Chir Scand 1944;83:1106.
29. Askew MJ, Mow VC. The Biomechanical Function of the
Collagen Ultrastructure of articular cartilage. J Biomech
Eng 1978;100:105115.
225
30. Spilker RL, Maxian TA. A mixed-penalty finite-element formulation of the linear biphasic theory for soft-tissues. Int J
Numer Methods Eng 1990;30:10631082.
31. van der Voet A. A comparison of finite element codes for the
solution of biphasic poroelastic problems. Proc Inst Mech Eng
[H] 1997;211:209211.
32. Wu JZ, Herzog W, Ronsky J. Modeling axi-symmetrical joint
contact with biphasic cartilage layersan asymptotic solution. J Biomech 1996;29:12631281.
33. Donzelli PS, Spilker RL, Ateshian GA, Mow VC. Contact
analysis of biphasic transversely isotropic cartilage layers
and correlations with tissue failure. J Biomech 1999;32:
10371047.
34. Wu JZ, Herzog W, Epstein M. Evaluation of the finite element
software ABAQUS for biomechanical modelling of biphasic
tissues. J Biomech 1998;31:165169.
35. Blankevoort L, Kuiper JH, Huiskes R, Grootenboer HJ.
Articular contact in a three-dimensional model of the knee.
J Biomech 1991;24:10191031.
36. Eberhardt AW, Keer LM, Lewis JL, Vithoontien V. An
analytical model of joint contact. J Biomech Eng 1990;112:
407413.
37. Hirokawa S. Three-dimensional mathematical model
analysis of the patellofemoral joint. J Biomech 1991;24:
659671.
38. Oloyede A, Flachsmann R, Broom ND. The dramatic influence of loading velocity on the compressive response of
articular cartilage. Connect Tissue Res 1992;27:211224.
39. Buschmann MD et al. Confined compression of articular
cartilage: linearity in ramp and sinusoidal tests and the
importance of interdigitation and incomplete confinement.
J Biomech 1998;31:171178.
40. Bursac PM, Obitz TW, Eisenberg SR, Stamenovic D. Confined and unconfined stress relaxation of cartilage: appropriateness of a transversely isotropic analysis. J Biomech
1999;32:11251130.
41. Haut Donahue TL, Hull ML, Rashid MM, Jacobs CR. How the
stiffness of meniscal attachments and meniscal material
properties affect tibio-femoral contact pressure computed
using a validated finite element model of the human knee
joint. J Biomech 2003;36:1934.
42. Beaugonin M, Haug E, Cesari D. Improvement of numerical
ankle/foot model: Modeling of deformable bone. Proc 1997
41st Stapp Car Crash Conf. Lake Buena Vista (FL): SAE,
Warrendale (PA); 1997. p 225237.
43. Beillas P, et al. Limb: Advanced FE model and new experimental data. Stapp Car Crash J 2001;45:469494.
44. Godest AC et al. Simulation of a knee joint replacement
during a gait cycle using explicit finite element analysis. J
Biomech 2002;35:267275.
45. Halloran J, Petrella A, Rullkoetter P. Explicit finite element
model predicts TKR mechanics. 49th Annu Meet, Orthopaedic Res Soc. New Orleans (LA); 2003. p 1312.
46. Benvenuti J-F. Mode lisation tridimensionnelle du genou
humain. Laboratoire de Ge nie Me dical. Lausanne, Switzerland:
EPF Lausanne; 1998.
47. Beillas P et al. Foot and ankle finite element modeling using
CT-scan data. 43rd Stapp Car Crash Conf. San Diego; 1999.
p 217.
48. Keyak JH, Skinner HB. Three-dimensional finite element
modelling of bone: Effects of element size. J Biomed Eng
1992;14:483489.
49. Ulrich D et al. The quality of trabecular bone evaluated with
micro-computed tomography, FEA and mechanical testing.
Stud Health Technol Inform 1997;40:97112.
50. Ulrich D, van Rietbergen B, Weinans H, Ruegsegger P. Finite
element analysis of trabecular bone structure: a comparison
226
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
of image-based meshing techniques. J Biomech 1998;31:
11871192.
Beillas P. Mode lisation des membres infe rieurs en situation
de choc automobile. Laboratoire de Biome canique. Paris,
cole Nationale Supe rieure dArts et Me tiers; 1999.
France: E
Noailles J. Mode lisation me canique par e le ments finis
de larticulation du genou. Laboratoire de Biome canique.
cole Nationale Supe rieure dArts et Me tiers;
Paris, France: E
1999.
Limbert GMT, Freeman MAR. Three dimensional finite element model of the human ACL. Simulation of a passive knee
flexion cycle. Analysis of deformations and stresses. 47th
Annu Meet Orthopaedic Res Soc. San Francisco (CA);
2001. p 794.
Hirokawa S, Tsuruno R. Hyper-elastic model analysis
of anterior cruciate ligament. Med Eng Phys 1997;19:637
651.
Butler DL et al. Location-dependent variations in the material properties of the anterior cruciate ligament. J Biomech
1992;25:511518.
Yamamoto K, Hirokawa S, Kawada T. Strain distribution in
the ligament using photoelasticity. A direct application to the
human ACL. Med Eng Phys 1998;20:161168.
Pioletti DP. Viscoelastic properties of soft tissues: Application
to knee ligaments and tendons. Departement de physique
Ecole polytechnique federale de Lausanne. Lausanne: Ecole
polytechnique federale de Lausanne; 1997.
Hirokawa S, Tsuruno R. Three-dimensional deformation
and stress distribution in an analytical/computational
model of the anterior cruciate ligament. J Biomech 2000;33:
10691077.
Taylor M, Tanner KE, Freeman MA, Yettram AL. Cancellous
bone stresses surrounding the femoral component of a hip
prosthesis: An elastic-plastic finite element analysis. Med
Eng Phys 1995;17:544550.
Aspden RM. A model for the function and failure of the
meniscus. Eng Med 1985;14:119122.
Anderst WJ, Tashman S. A method to estimate in vivo
dynamic articular surface interaction. J Biomech 2003;36:
12911299.
Walker PS, Hajek JV. The load-bearing area in the knee joint.
J Biomech 1972;5:581589.
Fukubayashi T, Kurosawa H. The contact area and pressure
distribution pattern of the knee. A study of normal and
osteoarthrotic knee joints. Acta Orthop Scand 1980;51:
871879.
Warner JJ. Articular contact patterns of the normal glenohumeral joint. J Shoulder Elbow Surg 1998;7:381388.
Scherrer PK, Hillberry BM, Van Sickle DC. Determining the
in vivo areas of contact in the canine shoulder. J Biomech Eng
1979;101:271278.
Soslowsky LJ et al. Quantitation of in situ contact areas at
the glenohumeral joint: A biomechanical study. J Orthop Res
1992;10:524534.
Dennis DA, Komistek RD, Hoff WA, Gabriel SM. In vivo knee
kinematics derived using an inverse perspective technique.
Clin Orthop Relat Res 1996: 107117.
Sheehan FT, Zajac FE, Drace JE. Using cine phase contrast
magnetic resonance imaging to non-invasively study in vivo
knee dynamics. J Biomech 1998;31:2126.
Papaioannou G, Tashman S, Nelson F. Morphology proportional differences in the medial and lateral compartment of
the distal femur. 50th Ann Meet, Orthopaedic Res Soc. San
Francisco, (CA); 2004. p 1256.
Treece GM, Prager RW, Gee AH. Regularised marching
tetrahedra: Improved iso-surface extraction. Comput Graph
1999;23:583598.
71. Papaioannou G, Tashman S. Validation of a lower limb model

based on 3D knee kinematics from a high speed biplane
dynamic radiography. In: Fotiadis DI, Dassios G, Kiriaki
K, Massalas CV, eds. Scattering Theory and Biomedical
Engineering Modelling and Applications. World Scientific;
2001.
72. Ateshian GA. A B-spline least-squares surface-fitting method
for articular surfaces of diarthrodial joints. J Biomech Eng
1993;115:366373.
73. Paul JP, Poulson J. The analysis of forces transmitted by
joints in the human body. 5th Int Conf Experimental Stress
Analysis. Udine, Italy; 1974.
74. Hefzy MS, Zoghi M, Jackson WT, DiDio LJA. Method to
measure the three-dimensional patello-femoral tracking.
Adv Bioeng 1988;8:4749.
75. Papaioannou G, Daly D, Spaepen A. Use of new optimalization tools in knee joint modelling. In: Dassios G, Fotiadis DI,
Kiriaki K, Massalas CV, eds. Scattering Theory and Biomedical Engineering Modelling and Applications. Singapore:
World Scientific; 2000. pp 282295.
76. Jonkers I, Spaepen A, Papaioannou G, Stewart C. An EMGbased, muscle driven forward simulation of single support
phase of gait. J Biomech 2002;35:609619.
77. Papaioannou G. A Three Dimensional Mathematical Model of
the Knee Joint. Bioengineering Ph.D. Thesis. Glasgow, (UK):
University of Strathclyde; 2000.
78. Van der Helm FC, Veeger HE, Pronk GM, Van der Woude
LH, Rozendal RH. Geometry parameters for musculoskeletal modelling of the shoulder system. J Biomech 1992;
25:129144.
79. Zajac FE. Muscle coordination of movement: a perspective. J
Biomech 1993;26(Suppl 1):109124.
80. Pandy MG, Berme N. A numerical method for simulating the
dynamics of human walking. J Biomech 1988;21:10431051.
81. Chao EY, Rim K. Application of optimization principles in
determining the applied moments in human leg joints during
gait. J Biomech 1973;6:497510.
82. Townsend MA, Tsai TC. Biomechanics and modelling of
bipedal climbing and descending. J Biomech 1976;9:227239.
83. Chou LS, Song SM, Draganich LF. Predicting the kinematics
and kinetics of gait based on the optimum trajectory of the
swing limb. J Biomech 1995;28:377385.
84. Pandy MG, Berme N. Quantitative assessment of gait determinants during single stance via a three-dimensional model
Part 2. Pathological gait. J Biomech 1989;22:725733.
85. Pandy MG, Berme N. Quantitative assessment of gait
determinants during single stance via a three-dimensional
modelPart 1. Normal gait. J Biomech 1989;22:717724.
86. Winter DA. The Biomechanics and Motor Control of Human
Gait. Waterloo: University of Waterloo Press; 1987.
87. Huang CY, Stankiewicz A, Ateshian GA, Mow VC. Anisotropy, inhomogeneity, and tension-compression nonlinearity
of human glenohumeral cartilage in finite deformation. J
Biomech 2005;38:799809.
88. Chan B, Donzelli PS, Spilker RL. A mixed-penalty biphasic
finite element formulation incorporating viscous fluids and
material interfaces. Ann Biomed Eng 2000;28:589597.
89. Zhang H, Totterman S, Perucchio R, Lerner AL. Magnetic
resonance image based 3D poroelastic finite element model of
tibio-menisco-femoral contact. Proc 23rd Annu Meet Am Soc
Biomechanics. Pittsburgh; 1999.
90. Perie D, Hobatho MC. In vivo determination of contact
areas and pressure of the femorotibial joint using non-linear
finite element analysis. Clin Biomech (Bristol, Avon) 1998;13:
394402.
91. Eckstein F et al. Quantitative relationships of normal cartilage volumes of the human knee jointassessment by
92.
93.
94.
95.
96.
97.
98.
99.
100.
101.
102.
103.
104.
105.
106.
107.
108.
109.
magnetic resonance imaging. Anat Embryol (Berlin) 1998;197:

383390.
Eckstein F, Reiser M, Englmeier KH, Putz R. In vivo morphometry and functional analysis of human articular cartilage with quantitative magnetic resonance imagingfrom
image to data, from data to theory. Anat Embryol (Berlin)
2001;203:147173.
Rudert MJ, et al. Articular cartilage thickness measurement
with MRI and multi-detector computed tomography (MDCT).
49th Annu Meet, Orthopaedic Res Soc. New Orleans (Lo);
2003. p 0571.
Bendjaballah MZ, Shirazi-Adl A, Zukor DJ. Biomechanics of
the human knee joint in compression: Reconstruction, mesh
generation and finite element analysis. Knee 1995;2:6979.
Xia Y. Magic-angle effect in magnetic resonance imaging of
articular cartilage: A review. Invest Radiol 2000;35:602621.
DiSilvestro MR, Zhu Q, Suh JK. Biphasic poroviscoelastic
simulation of the unconfined compression of articular cartilage: IIEffect of variable strain rates. J Biomech Eng
2001;123:198-200.
Eberhardt AW, Lewis JL, Keer LM. Normal contact of elastic
spheres with two elastic layers as a model of joint articulation. J Biomech Eng 1991;113:410417.
Donahue TL, Hull ML, Rashid MM, Jacobs CR. A finite
element model of the human knee joint for the study of
tibio-femoral contact. J Biomech Eng 2002;124:273280.
Weinans H, Sumner DR, Igloria R, Natarajan RN. Sensitivity
of periprosthetic stress-shielding to load and the bone density-modulus relationship in subject-specific finite element
models. J Biomech 2000;33:809817.
Papaioannou G, Yang K, Fyhrie D, Tashman S. Validation of
a subject specific finite element model of the human knee
developed for in-vivo tibio-femoral contact analysis. 50th
Annu Meet, Orthopaedic Res Soc. San Francisco; 2004. p
0358.
Couteau B et al. Finite element modelling of the vibrational
behaviour of the human femur using CT-based individualized geometrical and material properties. J Biomech
1998;31:383386.
Papaioannou G, Anderst W, Tashman S. Elevated joint contact forces in ACL-reconstructed knees: A finite element
analysis driven by in vivo kinematic data. IMECE0 03:2003
ASME Int Mech Eng Cong Exposition. Washington, DC;
2003.
Beillas P, Papaioannou G, Tashman S, Yang KH. A new
method to investigate in vivo knee behavior using a finite
element model of the lower limb. J Biomech 2004;37:1019
1030.
Li G, Lopez O. Reliability of a 3D finite element model
constructed using magnetic resonance images of a knee for
joint contact stress analysis. 23rd Proc Am Soc Biomech.
Pittsburgh; 1999.
Charras GT, Guldberg RE. Improving the local solution
accuracy of large-scale digital image-based finite element
analyses. J Biomech 2000;33:255259.
Dunbar WL, Jr., Un K, Donzelli PS, Spilker RL. An evaluation of three-dimensional diarthrodial joint contact using
penetration data and the finite element method. J Biomech
Eng 2001;123:333340.
Spilker RL, Donzelli PS, Mow VC. A transversely isotropic
biphasic finite element model of the meniscus. J Biomech
1992;25:10271045.
Morrison JB. The mechanics of the knee joint in relation to
normal walking. J Biomech 1970;3:5161.
Kettelkamp DB, Jacobs AW. Tibiofemoral contact area
determination and implications. J Bone Joint Surg Am
1972;54:349356.
227
110. Radin EL et al. Relationship between lower limb dynamics

and knee joint pain. J Orthop Res 1991;9:398405.
111. Collins JJ, Whittle MW. Impulsive forces during walking and
their clinical implications. Clin Biomech 1989;4:179187.
112. Tashman S, Leisen JC, Sherlitz C, Radin EL. Methods for the
reduction of heelstrike impulsive loading. Second World
Cong Biomech. Amsterdam, The Netherlands; 1994.
113. Rolf C et al. An experimental in vivo method for analysis
of local deformation on tibia, with simultaneous measures
of ground reaction forces, lower extremity muscle activity
and joint motion. Scand J Med Sci Sports 1997;7:144151.
114. Tashman S, Anderst W. Internal/external and varus/valgus
knee rotations are different in ACL-reconstructed and contralateral (intact) limbs during running. 49th Annu Meet,
Orthopaedic Res Soc. New Orleans (LA); 2003. p 0124.
115. Papaioannou G, Fyhrie D, Tashman S. Effects of
patient-specific cartilage geometry on contact pressure: An
in-vivo finite element model Of ACL reconstruction. 50th
Annu Meet, Orthopaedic Res Soc. San Francisco; 2004. p 1289.
116. Hollister SJ, Fyhrie DP, Jepsen KJ, Goldstein SA. Application of homogenization theory to the study of trabecular bone
mechanics. J Biomech 1991;24:825839.
117. Jacobs CR et al. NACOB presentation to ASB Young Scientist
Award: Postdoctoral. The impact of boundary conditions and
mesh size on the accuracy of cancellous bone tissue modulus
determination using large-scale finiteelement modeling.
North American Congress on Biomechanics. J Biomech
1999;32:11591164.
118. van Rietbergen B et al. Tissue stresses and strain in trabeculae of a canine proximal femur can be quantified from
computer reconstructions. J Biomech 1999;32:443451.
119. Yeni YN, Fyhrie DP. Finite element calculated uniaxial
apparent stiffness is a consistent predictor of uniaxial apparent strength in human vertebral cancellous bone tested with
different boundary conditions. J Biomech 2001;34: 16491654.
120. Blankevoort L, Huiskes R, de Lange A. The envelope of
passive knee joint motion. J Biomech 1988;21:705720.
121. Ahmed AM, Burke DL. In-vitro measurement of static
pressure distribution in synovial jointsPart I: Tibial surface of the knee. J Biomech Eng 1983;105:216225.
122. Besnault B, et al. A parametric finite element of the human
pelvis. 42nd Stapp Car Crash Conf Proc. Tempe (AZ); 1998. p
337.
123. Anderst W, Tashman S. A unique method to determine
dynamic in vivo articular surface interaction. 4th World Cong
Biomecha. Calgary, Alberta, Canada; 2002. p 520.
124. Anderst WJ, Les C, Tashman S. In vivo serial joint space
measurements during dynamic loading in a canine model of
osteoarthritis. Osteoarth Cartilage 2005;13(9):808816.
125. Demetropoulos CK. Dynamic evaluation of contact pressure
and effect of graft harvest at osteochondral donor site in the
knee, personal communication. 2003.
126. Kempson GE, Spivey CJ, Swanson SA, Freeman MA. Patterns of cartilage stiffness on normal and degenerate human
femoral heads. J Biomech 1971;4:597609.
127. Johnston RC, Smidt GL. Hip motion measurements for
selected activities of daily living. Clin Orthop Relat Res
1970;72:205215.
128. Bergmann G, Graichen F, Rohlmann A. Is staircase walking a
risk for the fixation of hip implants? J Biomech 1995;28:535553.
129. Andriacchi TP, Mikosz RP, Hampton SJ, Galante JO. Model
studies of the stiffness characteristics of the human knee
joint. J Biomech 1983;16:2329.
130. King AI. A review of biomechanical models. J Biomech Eng
1984;106:97104.
131. Cavanagh PR et al. The relationship of static foot structure
to dynamic foot function. J Biomech 1997;30:243250.
228
132. Matsen FA, Fu FH, Hawkins RJ. The Shoulder: A Balance of

Mobility and Stability. Rosemont: AAOS; 1992.
133. Chao EY, Morrey BF. Three-dimensional rotation of the
elbow. J Biomech 1978;11:5773.
134. Panjabi MM, Krag MH, Goel VK. A technique for measurement and description of three-dimensional six degree-offreedom motion of a body joint with an application to the
human spine. J Biomech 1981;14:447460.
135. Yang KH, Latouf BK, King AI. Computer simulation of
occupant neck response to airbag deployment in frontal
impacts. J Biomech Eng 1992;114:327331.
136. Walker PS. Human Joints and Their Artificial Replacements.

Springfield (IL): Thomas; 1977.
137. Youm Y, Yoon YS. Analytical development in investigation of
wrist kinematics. J Biomech 1979;12:613621.
See also CARTILAGE AND MENISCUS, PROPERTIES OF; HIP JOINTS, ARTIFICIAL;
HUMAN SPINE, BIOMECHANICS OF; LIGAMENT AND TENDON, PROPERTIES OF.
JOINT REPLACEMENT. See MATERIALS AND DESIGN FOR

ORTHOPEDIC DEVICES.
L
LAPARASCOPIC SURGERY. See MINIMALLY INVASIVE
The larynx has three important functions: protection of

the lower airways during swallowing; respiration; and
phonation.
Phonation is a complicated process in which sound is
produced for speech. During phonation, the vocal folds are
brought together near the center of the larynx by muscles
attached to the arytenoids. As air is forced through the
vocal folds, they vibrate and produce sound. The tone and
level of the sound can be changed by contracting or relaxing
the muscles of the arytenoids. As the sound produced by the
larynx travels through the throat and mouth, it is further
modified to produce speech.
Cancer of the larynx represented 0.7% of the total
cancer risk in 2001, and is the most common of all head and
neck cancers. However, head and neck cancers account for
only 9% of all cancers diagnosed annually.
Laryngeal cancer occurs about five times more frequently in males than females. It is rare prior to age 40,
after which the incidence in males increases rapidly with
age. Cigarette smoking is the most important cause of
laryngeal cancer, with smokers having a roughly 10-fold
higher risk than nonsmokers. Heavy alcohol consumption
also is a well-established risk factor.
Though laryngeal cancer is infrequent compared to
cancer of the breast, lung, and prostate, the literature
regarding this disease is substantial. This apparently disproportionate body of writing reflects the perceived importance of this neoplasm, which is in turn related to its
potential impact on peoples communicative ability: the
threat to a patients vocal organ is associated with profound
psychological and socioeconomic overtones.
Originally, larynx cancer treatment focused primarily
on cure by relentless and aggressive surgery. That era has
been followed by the emergence of conservative partial
laryngectomy, the development of more sophisticated
radiation methods, and organ-sparing strategies in which
chemotherapeutic, radiotherapeutic, and surgical techniques are used in a variety of combinations.
Total laryngectomy is one of the standard operations for
laryngeal carcinomas. The prognosis associated with laryngeal carcinoma has improved: As the curability of laryngeal carcinoma is now > 60%, many patients thus
survive for a long time after surgery.
Laryngectomy changes the anatomy: The lower respiratory tract is separated from the vocal tract and from the
upper digestive tract, and the laryngectomee breathes
through a tracheostoma. The direct connection between
the vocal tract and the upper digestive tract remains
unchanged.
Figure 2 shows the anatomic structure before laryngectomy (a) and after laryngectomy (b). Before laryngectomy
(Fig. 2a), air can travel from the lungs to the mouth (as
represented by the arrows), and the voice can be produced
and modulated. After the laryngectomy (Fig. 2b), air issuing from the tracheostoma cannot reach the mouth, and
sounds cannot be produced.
SURGERY.

GUIDO BELFORTE
MASSIMILIANA CARELLO
Politecnico di Torino
Torino, Italy
GUIDO BONGIOANNINI
MAURO MAGNANO
ENT Division Mauriziano
Hospital
Torino, Italy
INTRODUCTION
The larynx is a uniquely complicated organ strategically
located at the division of the upper aerodigestive tract into
the gastrointestinal tract and the airways. Alteration of its
function can have a significant impact on vocal, digestive,
and respiratory physiology (16). The hyoid bone, the
thyroid cartilage, and the cricoid cartilage form the outside
framework of the larynx. The mobile interior framework
consists of the leaf-shaped epiglottis and the arytenoid
cartilages. Each vocal cord stretches from an anterior
projection of the arytenoid to the anterior midline of the
inside thyroid cartilage. The arytenoid cartilages move in
both a rocking and sliding motion on the cricoid cartilage to
abduct and adduct the true vocal cords. The intrinsic muscles
of the larynx control the movement of the vocal cords.
A section of the larynx is shown in Fig. 1, which is a
posterior view of the cartilages and membranes (skeleton of
larynx).
Figure 1. Section of the larynx: posterior view of cartilages and

membranes.
229
230
Figure 2. Anatomy before (a) and after

(b) laryngectomy. (From Ref. 6.)
THE VOICE AFTER A LARYNGECTOMY

After laryngectomy, the patient is deprived of both the
vibrating sound source (the vocal folds) and the energy
source for voice production, as the air stream from the
lungs is no longer connected to the vocal tract (19).
For some laryngectomy patients, the loss of speech is
more important than survival itself. Consequently, a number of different methods for recovering phonation have
been developed, including: esophageal speech; artificial
larynx; and surgical laryngoplasty.
Numerous internal or external mechanical vibration
sources have been developed that cause the air in the vocal
tract to vibrate. These vibration sources can be powered by
air pressure (expired air from the tracheostoma in some
cases) or by an electric source with battery, and are thus
classified as pneumo-larynges or electrical larynges.
ESOPHAGEAL SPEECH
Rehabilitation of the laryngectomized patient is usually a
delayed process following recovery from surgery. After
surgery, some patients try aphonic lip speech enhanced
by buccal air trapping, while others choose written language as the method of communication.
Though most laryngectomized patients begin to learn
esophageal speech, this method of speech rehabilitation
requires a sequence of training sessions to develop the
ability to insufflate the esophagus by inhaling or injecting
air through coordinated muscle activity of the tongue,
cheeks, palate, and pharynx.
Patients are encouraged to attempt esophageal sound
soon after they are able to swallow food comfortably and
learn to produce esophageal sound by trapping air in the
mouth and forcing it into the esophagus. This produces a
burp-like tone that can be developed into the esophageal
voice.
There are various techniques for transporting air into
the esophagus.
With the injection technique, the tongue forces air back
into the pharynx and esophagus. This takes place in two
stages, with the tongue forcing the air from the mouth back
into the pharynx in the first stage, and the back of the
tongue propelling the air into the esophagus in the second
stage. For air to be transported into the esophagus, it is
extremely important that these two stages be correctly
synchronized.
With the inhalation method of esophageal speech, the
patient creates a pressure in the esophagus that is lower
than atmospheric pressure. As a result of this pressure
difference, air will flow through the mouth past the upper
segment of the esophagus into the lower esophagus. The
patient will need to inhale air to be able to create a low
endothoracic and esophageal pressure.
The last technique of capturing air is by swallowing air
into the stomach.
Voluntary air release or regurgitation of small
volumes vibrates the cervical esophageal inlet, hypopharyngeal mucosa, and other portions of the upper aerodigestive tract to produce a burp-like sound. Articulation by
the lips, teeth, palate, and tongue produces intelligible
speech.
Esophageal speech training is time consuming, frustrating, and sometimes ineffective.
Its main disadvantage is the low success rate in acquiring useful voice production, which varies from 26 to 55%. In
addition, esophageal speech results in low-pitched (60
80 Hz) and low intensity speech, whose intelligibility is
often poor. Age is the most important factor in determining
success or failure: older patients are less successful in
learning esophageal speech.
The airway used to create the esophageal voice is shown
in Fig. 3. Direction of flow is indicated by the arrows,
making it possible to distinguish between the pulmonary
air used for breathing and the mouth air used for speech.
THE ELECTRONIC ARTIFICIAL LARYNX
Wright introduced the first electrolarynx in 1942. The most
widely used electronic artificial larynx is the handheld
transcervical device, or electrolarynx. This electrical device
contains a vibrating diaphragm, which is held against the
throat and activated by a button to inject vibratory energy
through the skin and into the hypopharynx. By mouthing
231
Figure 5. Examples of commercial electrolarynx. (From Ref. 5.)

Figure 3. Oesophageal speech. (From Ref. 6.)
words, the laryngectomee converts the vibrations into a

new, low frequency voice for speech.
An example of electrolarynx application is shown in
Fig. 4, where it is possible to distinguish the airway from
the sound way and the positioning of the device in contact
with the neck. Examples of commercial electrolarynges are
shown in Fig. 5.
The same operating principle has been used in recent
years to develop a transoral device (Fig. 6), which is placed
in the mouth, where it is housed in an upper denture or an
orthodontic retainer. The system consists of a control
circuit, a loud speaker, and rechargeable batteries positioned inside the denture, as well as a charging port so that
the batteries can be recharged outside the mouth.
FROM THE PNEUMOLARYNX TO THE VOICE PROSTHESIS
The artificial larynx has undergone many transformations
over the years, and continues to do so today. The first types
Figure 4. Electrolarynx speech. (From Ref. 6.)
of pneumolarynges, which included neck or mouth types

and internal or external types were developed in 1860,
when the first attempts at voice rehabilitation through
surgery or prosthetization were made. A device was used
to direct air from the lungs via a small tube to the mouth
(1,3,5,7,9).
Experimental research started in 1860 with Ozermack
and Burns, though it was not until 1874 that Billroth used
an internal prosthesis designed by Gussenbauer (Fig. 7).
This device was fitted in the trachea via the stoma with a
cannula featuring a branch that entered the oral cavity.
Other similar prostheses were developed and used
(Gosten, Gluck, etc.). Results, however, were not particularly satisfactory, as these devices were plagued by problems, such as tissue necrosis and leakage of food and
liquids into the windpipe and lungs, thus causing infections (i.e., pneumonia).
Figure 8 shows the external prosthesis developed by
Caselli in 1876. Figure 9 shows the application of the
external prosthesis developed by Briani in 1950.
Figure 6. Transoral device. (From Ref. 5.)
232
Figure 10. Example of removable prosthesis (Bivona).
Figure 7. The Gussembauer prosthesis. (From Ref. 3.)
Figure 8. The Caselli prosthesis. (From Ref. 3.)
Figure 9. The Briani prosthesis. (From Ref. 3.)
Since 1960, surgical technique has been improved and a

number of different types of prosthesis have been invented
that take the importance of reproducing a voice that is
similar to the original voice into account.
Tracheoesophageal puncture (TEP) or fistula with a voice
prosthesis is the most popular of the new techniques, and has
become the standard method of voice restoration after laryngectomy. However, it is usable only for selected patients.
Singer and Blom introduced the first TEP in 1980 (7,8).
The aim was to create a permanent fistula (puncture or shunt)
into the posterior tracheoesophageal wall between the trachea and the esophagus tract, so the pulmonary air can shunt
from the airway into and up the esophagus. In this way, the
vibration of the anatomic structures produces a noise.
A removable or fixed one-way type prosthesis can be
positioned in the fistula.
The removable, or nonindwelling, prosthesis (Bivona,
Blom-Singer duckbill) can be removed daily for cleaning by
the patient, and is taped to the peritracheostomal skin.
Figure 10 shows a photo of two Bivona valves; lengths
differ in order to adapt the prosthesis to the patient.
The fixed, or indwelling, prosthesis cannot be removed
daily, but is surgically placed in the fistula under local
anesthesia. The operating principle of this type of prosthesis (known as a phonatory valve or voice button) can be
explained with reference to Fig. 11.
Pulmonary air can be pushed through the valve into the
esophagus for speech during expiration in two ways: by the
laryngectomee, who covers the tracheal stoma with a finger
(bottom left, Fig. 11) or automatically by a tracheostoma
breathing valve (bottom right, Fig. 11).
Figure 11. Speech with phonatory prosthesis. (From Ref. 6.)
The tracheostoma breathing valve contains an elastic

diaphragm installed in a peristomal housing, which permits
normal respiration during silent periods. Expiratory air for
speech shuts off the pressure-sensitive diaphragm, and is thus
diverted through the valve into the esophagus. This device
eliminates the need for manual stoma occlusion during speech.
In both cases, air from the lung reaches the esophagus
and causes the mucosal tissue tovibrate. The resulting
sound can be modulated by the mouth, teeth, oral cavity,
and so on, to produce the new voice.
Most speakers that have prosthesis do not have difficulties with articulation, rate of speech, or phonatory duration.
If esophageal speech depends on gulping or trapping air
using the phonatory valve, the resulting speech depends on
expiratory capacity. Voice quality is very good, and may
resemble the original voice.
Poor digital occlusion of the tracheostoma as well as
poor tracheostoma valve adherence allows pulmonary air
to escape from the stoma prior to its diversion through the
prosthesis into the oesophagus for voice production. In fact,
the bleeding off of pulmonary air limits phonatory duration
and, therefore, the number of words that can be spoken.
The one-way valve design of the prosthesis prevents
aspiration (of food and liquid) from the esophagus to the
trachea. An example of a phonatory valve is shown in
Fig. 12 (11,12). The prosthesis illustrated in this sketch
consists of an air-flow tracheal entry, whereby an endo-
Figure 13. Examples of fixed commercial prosthesis. (Staffieri,

Groningen standard, Groningen low pressure, Panje, Provox).
tracheal retention collar is connected to the mucosa (or

tracheal flange), a hollow cylindrical tube (whose length
depends on the patients physical characteristics) connecting the trachea to the esophagus, an endoesophageal
flange, and a dome (or hat) that closes the proximal endoesophageal end of the tube. Via the razor-thin slit (or
esophagus exit), the hat enables airflow to pass from the
trachea to the esophagus when there is a positive differential pressure, and prevents the reverse flow of liquid (or
food) when the differential pressure becomes negative. The
arrows represent the airflow paths.
Hat shape and the extension of the razor-thin slit can
differ according to valve type. The razor-thin slit may be
located at the base of the hat (Staffieri, Groningen low
pressure), at the center of the hat (Panje, Groningen
standard), or inside the hollow tube (Provox, Blom-Singer).
Though valve geometry and shape may vary, the operating principle remains the same.
Several commercial prostheses are shown in Fig. 13:
from left to right, they include the Staffieri, Groningen
standard, Groningen low pressure, Panje, and Provox types.
Fixed and removable prostheses are available in different lengths, which usually range from 6 to 12 mm to enable
the surgeon to select the dimensions that are best suited to
the patients physical characteristics, for example, posterior tracheoesophageal wall thickness.
To compare valve performance in the laboratory, most
authors use valve airflow resistance (7,11), which is defined
as the ratio of pressure to flow-rate. Figure 14 show an
example of resistance versus flow-rate characteristics
obtained with experimental tests on commercial valves (11).
Low resistance allows air to pass freely though the
prosthesis with little effort on the part of the patient, and
Figure 12. Sketch of phonatory valve or prostheses.
80
Staffieri
Resistance [kPa/(dm3/s)]
70
Provox
60
50
Groningen
standard
40
Groningen low
pressure
Panje
30
20
10
0
0
0,05
0,1
0,15
0,2
Flow-rate [dm3/s (ANR)]
0,25
0,3
233
0,35
Figure 14. Resistance of commercial valves.
234
LENSES, INTRAOCULAR
is thus the most favorable condition. Different materials

have been used for phonatory prostheses. For indwelling
prostheses, silicone rubber is the material of choice, though
other materials such as polyurethane have also been used.
The most significant problem affecting voice prostheses is
the formation of a thick biofilm on the esophageal surface, as
the esophageal environmental around the prosthesis provides ideal growing conditions for bacteria, fungi, and yeast.
In this area, secretions from the trachea, the mucous
membranes in the oropharynx and esophagus, and saliva
from the oral cavity create an optimal pabulum for
microorganisms, which can thus adhere to the prosthesis
surface. Though biofilm does not lead immediately to valve,
malfunction, colonies growing around the prosthesis can
increase airflow resistance, blocking the valve or preventing it from closing correctly. This causes saliva to leak from
the oral cavity through the esophagus lumen. The biofilm
also grows into the valve material (silicone rubber).
Massive microorganism growth is more frequent in
indwelling prostheses than in non-indwelling types, as
the latter are regularly removed and cleaned.
The first step of this colonization is the formation of a
very thin layer of organic origin, called a conditioning film.
Microorganisms adhere to the thin layer of saliva conditioning film on the inner esophageal surface of the device
and form the first stratum of biofilm, which is nearly
impossible to eradicate. There are few methods for reducing the growth of bacteria and fungi: One possibility is to
apply (through oral instillations) antimycotic drugs that
reduce biofilm formation and increase prosthesis life
(1,2,7,8,13,14). The valve has a limited life, varying from
4 to 8 months. After this period, the valve must be
removed and changed in an outpatient procedure because
of a dysfunctional mechanism caused by the biofilm, which
causes food and liquids to leak from the esophageal area to
the trachea.
Figure 15 shows a new Provox valve (left) and a Provox
valve after eight months of use by a patient (right). Silicon
rubber deterioration caused by microbial action and adherent deposits of microorganisms is clearly apparent.
Current work with voice prostheses focuses on improving their aerodynamic characteristics (low airflow resistance) and developing more resistant materials that can
extend the life of the device. For the patient, in any case,
the most important thing after a total laryngectomy is to be
hopeful: a prosthesis can provide a new voice, improving
the quality of life.
BIBLIOGRAPHY
1. Mahieu HF. Voice and speech rehabilitation following laryngectomy. Ph.D. dissertation, Rijksuniversiteit Groningen; 1988.
2. Pighi GP, Cristoferi V. Protesi tacheo-esofagee. Bologna:
Arianna Editrice; 1998.
3. Societe Francaise dOto-Rhino-Laryngologie et de Pathologie
cervico-faciale. Re habilitation de la voix et de la de glutition
apre`s chirurgie du larynx. Arnette, Paris; 1992.
4. Testut L, Jacob O. Trattato di anatomia topografica. Vol. II
Collo - Torace Addome, Utet, Torino, 1998.
5. Harrison LB, Sessions RB, Hong WK. Head and neck cancer.
Lippinocott Williams & Wilkins; 2004; 178189.
6. http://www.orl.nl
7. Blom ED. Tracheoesophageal voice restoration: origin, evolution,
state of the art. Folia Phonatrica Logopaedica 2000; 52:1423.
8. Blom ED, Singer MI, Hamaker RC. Tracheoesophageal voice
restoration following total laryngectomy. San Diego: Singular
Publishing Group Inc.; 1998.
9. Brown DH, et al. Postlaryngectomy voice rehabilitation: state
of the art at the millennium. World J Surg 2003;27(7):824831.
10. Nakamura T, Shimizu Y. Thacheal, laryngeal and esophageal
replacement device. The Biomedical Engineering Handbook
CRC and IEEE Press; 2000, p 136/1136/13.
11. Belforte G, Carello M, Miani C, Staffieri A. Staffieri tracheooesophageal prosthesis for voice rehabilitation after laryngectomy: an evaluation of characteristics. Med Biol Eng Comput
1998;36:754760.
12. Staffieri M, Staffieri A. A new voice button for post-total
laryngectomy speech rehabilitation. Laryngoscope 1988;
98(9):10271029.
13. Schwandt LQ, et al. Prevention of biofilm formation by dairy
products and N-acetylcysteine on voice prostheses in an artificial throat. Acta Otolaryngol 2004;124:726731.
14. Leunisse C, et al. Biofilm formation and design features of
indwelling silicone rubber tracheoesophageal voice prosthesesAn electron microscopical study. J Biomed Mater
Res 2001;58(5):556563.
See also COMMUNICATION
DEVICES; PULMONARY PHYSIOLOGY.
LASER SURGERY. See ELECTROSURGICAL UNIT (ESU).

LASERS, IN MEDICINE. See FIBER OPTICS IN MEDICINE.
LENSES, CONTACT.
See CONTACT
LENSES.
LENSES, INTRAOCULAR
HABIB HAMAM
Universite de Moncton
Moncton, New Brunswick,
Canada
INTRODUCTION
Figure 15. Provox valve: (a) new and (b) old.
Ridleys implantation (1949) of the first intraocular lens

(IOL) marked the beginning of a major change in the
practice of ophthalmology. The IOLs are microlenses
placed inside the human eye to correct cataracts, nearsightedness, farsightedness, astigmatism, or presbyopia.
There are two types of IOLs: anterior chamber lenses,
LENSES, INTRAOCULAR
235
which are placed in the anterior chamber of the eye

between the iris and the cornea, and posterior chamber
IOLs, which are placed in the posterior chamber behind the
iris and rest against the capsular bag. Procedures for
implanting the IOLs and technologies for manufacturing
them in various sizes, thicknesses, and forms as well as
with various materials progressed tremendously in the last
decade. Multifocal IOLs are one of the important signs of
this progress. While monofocal IOLs, the most commonly
used, are designed to provide clear vision at one focal
distance, the design of multiple optic (multifocal) IOLs
aims to allow good vision at a range of distances.
INTRAOCULAR LENSES: WHAT AND WHY?

An intraocular lens, commonly called IOL, is a tiny artificial lens implanted in the eye. It usually replaces the
faulty (cataractous) cristalline lens. The most common
defect of the natural lens is the cataract, when this optical
element becomes clouded over. Prior to the development of
IOLs, cataract patients were forced to wear thick coke
bottle glasses or contact lenses after the surgery. They
were essentially blind without their glasses. In addition to
IOLs replacing the crystalline lenses, a new family of IOLs,
generally referred to as phakic lenses, is nowadays subject
of active research and development. (Phakos is the Greek
word for lens. Phakic is the medical term for individuals
who have a natural crystalline lens. In Phakic IOL surgery,
an intraocular lens is inserted into the eye without removal
of the natural crystalline lens.) These IOLs are placed in
the eye without removing the natural lens, as is completed
in cataract surgery. They are used to correct high levels of
nearsightedness (myopia) or farsightedness (hyperopia).
An IOL usually consists of a plastic lens with plastic side
struts called haptics to hold the lens in place within the
capsular bag. The insertion of the IOL can be done under
local anesthesia with the patient awake throughout the
operation, which usually takes <30 min in the hands of an
experienced ophthalmologic surgeon (Fig. 1).
Figure 1. Implantation of an IOL: Since the natural lens is left

undistrubed, the operation is much simpler than a cataract operation.
The entire procedure consists of making a small incision at the edge of
the cornea and placing the appropriate tiny plastic lens in the space
between the iris and the cornea, (the anterior chamber). Stitches are
used to close the incision.
Spitfire planes. He then concluded the poly(methyl methacrylate) (PMMA) material of the canopies (windshield) was
compatible with eye tissue (4). This observation sparked
the idea for inserting an artificial lens in the eye. Ridley,
who was convinced this lens should be placed in the posterior chamber, designed a disk-shaped lens, much like the
natural lens, with a small peripheral flange allowing him to
hold the lens with forceps (4). The artificial lens, made
entirely of PMMA, weighed slightly > 100 mg in air and
was 8.35 mm in diameter. In several cases, he attempted
to implant the lens following intracapsular surgery using
the vitreous base for support (5). On November 29, 1949,
the first successful IOL implantation was performed at
St. Thomas Hospital in London (6,7). While far from
perfect, the procedure worked well enough to encourage
HISTORICAL OVERVIEW
The idea of the IOL dates back to the beginning of modern
cataract surgery when Barraquer developed keratomileusis (1). However, the first implantation of an artificial lens
in the eye was probably attempted in 1795 (2). References
to the idea of the IOL before World War II in ophthalmic
literature are rare. There has been mention of limited
animal experiments using both quartz and plastic material
performed in the 1940s, but nothing had come of these
efforts (3).
The most important step toward the implantation of
IOLs came as a result of World War II pilots, and the
injuries sustained when bullets would strike the plastic
canopy of their aircraft (Fig. 2), causing small shards of
plastic to go into their eye. In the late 1940s, Howard Ridley
was an RAF ophthalmologist looking after these unfortunate pilots and observed, to his amazement, little or no
reaction in cases in which the material had come from
Figure 2. Invention of the IOL: During World War II, Fighter

pilots were sometimes involved in accidents where the plastic
windshield (canopy) of their aircraft was shattered. Doctors
found that fragments of the canopy that entered the eye were
tolerated by the eye tissues. They might remain inside the eye for
years, and the eye would not react.
236
LENSES, INTRAOCULAR
further refinement. Then, over a decade, Ridley implanted

several hundred IOLs (8).
Though Ridley was ahead of his time, his method was
subject to serious criticism. Complications were common
and failure rates > 50% were often contemporaneously
quoted. Fixation was dependent on the formation of adhesions between the iris and the capsule. Several ophthalmologists strongly opposed to his procedure. Implantation
in the anterior chamber was technically easier and was
compatible with intracapsular surgery. Also, fixation could
be achieved at the time of implantation by adding haptic
struts to the lens that could be wedged into the angle. The
first anterior chamber lens was implanted by Baron in 1952
(8).
To make intraocular lens implantation safe, developments in lens design and surgical techniques were
required. Lens implantation did not become widely adopted
as an integral part of cataract surgery until the 1980s. Key
advances were the introduction of viscoelastic fluids to
protect the cornea during implantation and flexible haptics
to enhance long term stability of the IOL (9).
With traditional single vision monofocal IOLs, people
generally experience good vision after surgery at a single
focal point, either near or at a distance. The multifocal IOL
(10) was designed in the mid-1990s to provide a full range
of vision with independence from glasses in most situations.
Besides, the invention of phakic lenses is no less important than Ridleys invention. These IOLs were introduced
by Strampelli (11) and later popularized by Barraquer in
the late 1950s (12). Phakic IOLs are becoming more popular because of their good refractive and visual results and
because they are easy to implant in most cases (13). In the
beginning, the design was a biconcave angle-supported
lens. These lenses were abandoned following serious angleand endothelium-related complications. By the late 1980s,
Baikoff (14,15) introduced a myopia lens with Kelman-type
haptics (16). This design had many problems, leading to its
design modification a number of times. Fyodorov, the
inventor of radial keratotomy (17), introduced the concept
of a soft phakic lens in the space between the iris and the
anterior surface of the crystalline lens (18).
Based on earlier works of Worst, winner of the Binkhorst Award for innovation in ophthalmology in 1976,
Fechner introduced phakic myopia lens of iris claw design
in 1986 (19). This IOL is then referred to as WorstFechner
lens (20). Many companies around the world manufactured
it in various models. Today, people usually identify it by the
name of Artisan IOL.
MATERIAL OF THE IOL

Many factors, such as the optical quality of the system of
the eye (aberrations, . . .), presence of inflammation, cost, and
wound size, depend on the material and the form of the IOL.
From the point of view of flexibility, there are two
families of IOLs: foldable and inflexible lenses. Foldable
IOLs are generally made of acrylic or silicone. They can be
rolled up and inserted through a tube with a very small
incision not requiring any stitches. Inflexible IOLs, typically made of PMMA, require a larger incision because they
are unfoldable. Most lenses are biconvex, thus optically
equivalent upside down. However, most lenses have haptics which are generally angled to push the posterior optics.
Four basic materials are used for IOLs: PMMA, silicone,
acrylic and collamer. Other materials are also used. For
example, some manufacturers replace silicon by a hydrophilic biocompatible polymer, called collamer. Many IOLs
have been made from PMMA plastic, the same plastic the
original hard contact lenses were made of. Silicon IOLs are
foldable. Folding an IOL allows it to be inserted through a
smaller incision. A smaller incision heals faster and
induces less postop astigmatism. Some foldable lenses
are now being made of acrylic plastic. While acrylic and
silicone lenses are very popular, PMMA is the time-tested
material but, as stated above, requires a large incision.
Silicone oil is also a problem for silicone IOLs in that the
view of the fundus can be severely degraded. This is less so
for PMMA and hydrophobic acrylic IOLs and least for
hydrophilic acrylic. Although this is a relative contraindication for silicone IOLs in the face of significant vitreoretinopathy, a solvent exists that eliminates the problem
(21). Collamer is a new hydrophilic material just recently
released that has shown some interesting properties. It has
been shown to exhibit less internal reflectance than other
lens materials including silicone, acrylic, and PMMA (22).
It reduces the risk of long-term inflammation (23). Table 1
summarizes the characteristics of the four materials.
VARIOUS PARAMETERS FOR IOLs

Are age, race, and sex important parameters for IOL
implantation? Age at which surgery is performed turned
out to be of great importance (2428). The ideal age should
be at around 18 years when the refraction stabilizes.
However, in specific circumstances, in the interest of the
minor patient, the parents and the surgeon can opt to
perform phakic lens implantation at an earlier age. Studies
Table 1. Four Commonly Used IOLs Materials and their Advantages and Drawbacks
Material
Flexibility
Advantages
Drawbacks
PMMA
Rigid
larger incision, not foldable
Silicone
Foldable
Low cost, less inflammation, long-term experience,

good bicompatibility
Smaller incision, injectable
Collamer
Acrylic
Foldable
Foldable
Smaller incision, less inflammation, very good bicompatibility

Smaller incision, less inflammation, high refraction
index (thin IOL), good bicompatibility
high cost, more inflammation,

cannot use with silicon oil
high cost, short term experience
high cost
LENSES, INTRAOCULAR
of the suitable age of IOL implantation in children have

been carried out (24). A 3-year-old child has been qualified
with IOL implantation, the child younger than 9 years old
should be implanted with a normal adult IOL and then
corrected with glasses, and a child after 10 years old should
be directly implanted with a proper dioptric IOL (24). Some
researchers evaluated the influence of cataract surgery on
the eyes of children between 1 and 5 years old. They
concluded that cataract surgery, either extraction with
or without IOL implantation, did not retard axial elongation in children above 1 year old (25). Comparisons between
children with congenital or developmental lens opacities
who underwent extracapsular cataract extraction and children with normal eyes have been carried out (26). The
pattern of axial elongation and corneal flattening was
similar in the congenital and developmental groups to that
observed in normal eyes. No significant retardation or
acceleration of axial growth was found in the eyes
implanted with IOLs compared with normal eyes. A myopic
shift was seen particularly in eyes operated on at 48 weeks
of age and it is recommended that these eyes are made 6 D
hyperopic initially with the residual refractive error being
corrected with spectacles (26).
To our knowledge, IOL implantation does not depend on
race and sex.
OPTICAL QUALITY
Two optical qualities are distinguinshed: the intrinsic
optical quality of the IOL and the optical quality of the
system of the eye including the IOL. Many factors, such as
the material and the geometrical profile of the IOL, influence the intrinsic quality of this optical element. Axial shift
(decentration), transversal rotation (not around the optical
axis), and deformation (mechanical stresses, . . .) of the IOL
are examples of factors affecting the optical quality of the
whole system of the eye even in case the IOL alone is a
perfect optical element. Thus, the optical quality of the IOL
is certainly important. However, for vision, the determinant factor is the optical quality of the whole system of the
eye in which the IOL is implanted. Several studies have
been undertaken to assess the optical quality of the IOL
and the optical quality of the whole system of the eye.
Before progressing in this section, let us briefly introduce
the notion of optical quality.
Aberrations
Stated in wave optics, the system of the eye should transform the input wavefront into a perfect convergent spherical wavefront that has the image point as center (2931).
Note that an optical wavefront represents a continuous
surface composed of points of equal phase. Thus all imageforming rays, which travel normal to the exit spherical
wavefront, meet in the focal point in phase, resulting in
maximum radiant energy being delivered to that point. In
reality, this situation never occurs. The rays modified by
the optical system do not converge entirely to a common
point image. For one object point correspond several image
points that form a blurred image. This deviation from the
ideal case is called optical aberration, or merely aberration,
237
and is a measure of the optical quality of the system.

Aberration can be quantified either with respect to the
expected image point or to the wavefront corresponding to
this ideal point. If the real output wavefront is compared to
the ideal one, it is called the difference between them
wavefront aberration (29). All human eyes suffer from
optical aberrations that limit the quality of the retinal
image (3236). Several metrics have been proposed to
measure the optical quality of the system of the eye
(3741). Let us return back to IOLs now. Optical quality
of multifocal IOLs will be treated in the section devoted to
this kind of lens.
Optical Quality of the IOL
The optical quality of IOL was the subject of intensive
studies. Several common but some contrasted results have
been obtained. An exhaustive study goes beyond the scope of
this document. We limit our attention to some recent results.
Tognetto et al. (42) evaluated the optical quality of different
IOLs by using an optical test bench. The purpose of the study
was to evaluate the optical quality of IOLs and not to evaluate
the optical performance of these lenses once implanted.
Three randomly acquired samples of 24 different models of
foldable IOLs were compared. The conclusion is that different
IOLs can transmit different spectra of spatial frequencies.
The best frequency response was provided by acrylic IOLs,
particularly those with an asymmetrically biconvex profile.
This could be due to a reduction of optical degradation
provided by this type of profile. A lens with a higher frequency
response should create a better quality of vision once
implanted, and the frequency response should therefore be
considered when choosing the intraocular lens model (42).
Negishi et al. (43) evaluated the effect of chromatic
aberrations in pseudophakic eyes with various types of
IOLs. Their results show that longitudinal chromatic aberrations of some IOLs may degrade the quality of the retinal
image. They concluded that attention must be paid to the
detailed optical performance of IOL materials to achieve
good visual function.
In a comparative study (44), Martin found that the
collamer IOL reduced the number of induced higher order
aberrations when compared with acrylic and silicone lenses.
Indeed, he found that the collamer IOL has 55117%
fewer induced higher order aberrations than acrylic or
silicone materials. As a consequence, it produces less
postop glare. He concluded the collamer lens provides
clearer vision than the other lenses.
Optical Quality of ARTISAN Lenses
Brunette et al. (45) evaluated the optical quality of the eye
before and after the insertion of an ARTISAN phakic
intraocular lens for the treatment of high myopia (range
20.50 to 9.75D). Consecutive patients implanted with
the ARTISAN lens by a single surgeon were prospectively
enrolled. One eye per subject was tested. The wavefront
aberration was calculated from images recorded with a
Hartmann-Shack sensor (46,47). The PSF and the MTF
were also computed from the wavefront aberration. It was
concluded that preliminary data using the Hartmann
Shack wavefront sensor have not revealed a tendency
238
LENSES, INTRAOCULAR
toward deterioration of the optical performance following

the insertion of an ARTISAN lens for the treatment of high
myopia. The HartmannShack sensor is a useful tool for
the objective assessment of the image optical quality of eyes
with a phakic intraocular lens.
MULTIFOCAL IOLs
Unlike the natural lens, the curvature of current intraocular lenses cannot be changed by the eye. Standard
intraocular lenses provide good distance vision, and the
patient needs reading glasses for near vision. Newer
bifocal intraocular lenses give distance vision in one area
and near vision in another area of the vision field. How
does it work?
The basic idea consists in providing a lens with two
posterior focal points instead of one. The IOL is no longer
monofocal. It becomes bifocal. Two solutions are possible:
refractive or diffractive bifocal lens. A third solution consists in combining both approachs together.
Diffractive Lenses
The idea comes from the principle of the Fresnel zone plate.
It consists in designing a binary diffractive phase element
so when the incident wave comes across this diffractive
element, all the resulting waves, coming from all points of
the zone plate, arrive in phase at a certain (focal) point.
They then superimpose constructively, yielding a focusing
behavior. As shown in Fig. 3, waves traveling along various
segments arrive in phase at the focal point since the optical
paths differ by a multiple of the wavelength. To fulfill this
condition, the thickness d is chosen so that it introduces a
phase shift of p: d 2k 1 l=2n 1 (optical path:
2k 1 l=2). In Fig. 3, k 0, yielding d l=2n 1,
where n is the refraction index of the diffractive lens.
The radii of the rings (Fig. 3) verify the following rule:
Figure 3. Binary diffractive phase element: Fresnel zone plate.

Waves traveling along various segments arrive in phase at the
focal point since the optical paths differ by a multiple of the
wavelength. To fulfill this condition, the thickness d is chosen
so that it introduces a phase shift of p: d (2k + 1) l/[2(n1)]
(optical path: (2k + 1) l/2): In the figure, k 0.
p
p
rm m
r1 ml f , where r1 is the radius of the smallest
circle and f is the focal length. Without grooves on the
diffractive, the waves traveling along the segments, represented by dashed lines in Fig. 3, would arrive in phase
opposition with respect to other waves. In reality, the
binary form (two phase levels) of the diffractive lens does
not fully ensure the condition of coming in phase at the
focal point. For rigor, several phase levels are required
(Fig. 4). In general, the diffraction efficiency h, which is
defined as the ratio of the focused energy to the incident
energy, increases with the number of phase levels L according to the following formula (48): h sin2 p=L=p=L2 .
Using two phase levels (binary), only 41% of the energy
is focused. The rest is scattered in space. A four level
diffractive lens focuses 81% of the input energy (it scatters
only 19% of the incident energy).
To obtain a bifocal diffractive lens, we need to focus rays
on two focal points at distances f1 and f2. It can be done by
providing two series of zones (rings). The first
series, the
p
1
inner one, involves radii verifying rm ml f1 (with
m 1,2,. . ., M), whereas the radii in the second
p series,
2
the outer one, satisfy the condition r p pl f2 (with
p M1,. . ., P). To obtain a multifocal diffractive lens,
we need more series of radii.
Refractive Lenses
An alternative to obtain a multifocal length consists in
modifying the surface profil of a conventional biconvex lens
so that it provides two or more different focal points for
light convergence. The technique consists in designing a
spherical refractive surface that has additional refracting
surfaces to give a near add (Fig. 5), or a near and intermediate add. The principle of multifocal refractive lenses is
illustrated in Fig. 6a. Refractive IOLs with several focal
points are commercialized in various models. For example,
one of the models includes five refractive zones targeting
distance, intermediate and near vision (Fig. 6b). The IOL
uses continuous aspheric optics to ensure that 100% of the
light entering the eye reaches the retina. The lens uses five
concentric zones with the first, third, and fifth zones being
far dominant and second and fourth zones being near
dominant (49). The light distribution is arranged so that
50% of light is distant focussed, 13% is focussed for intermediate vision and 37% for near vision. The near add
Figure 4. Bifocal hybrid refractive/diffractive IOL: Anterior

surface broken up into a refractive zone and a second zone
composed of concentric diffractive rings.
LENSES, INTRAOCULAR
239
reveal comparable properties for distance vision and a

superiority of the bifocal diffractive lens over the refractive
multifocal lens for near vision. This mean due to the fact
that the incoming light is distributed over different zones
in the refractive lenses.
Hybrid Lenses
Figure 5. Bifocal refractive IOL: Spherical refractive surface that

has additional refracting surfaces to give a near add.
comprises of 3.5 dioptre intraocular power equivalent to a

2.752.85 add in the spectacle plane (49).
Optical Quality of Refractive and Diffractive Lenses
This discussion will be limited to some recent results. Pieh
et al. (50) compared the optical properties of bifocal diffractive and multifocal refractive intraocular lenses. The
IOLs were manufactured by different companies. A model
eye with a pupil 4.5 mm in diameter was used to determine
the point spread function (PSF) (30,51) of the distance focus
and near focus of the IOLs to compare them with PSFs of
foci of corresponding monofocal lenses. For interpreting the
PSFs the through focus response, the modulation transfer
function (MTF) (51), and the Strehl ratio (51) were evaluated. They concluded the modulation transfer functions
Figure 6. Multifocal refractive IOLs: (a) several refractive

surfaces with different curvatures. Each one provides a focal point
(b) a commercial zonal progressive IOL with five concentric zones
on the anterior surface. Each zone repeats the entire refractive
sequence corresponding to distance, intermediate and near vision,
resulting in wide-range vision.
Diffractive and refractive optics (Fig. 4) can be combined.

In this type of lens, the basic spherical refractive surface is
broken up into a refractive zone and a second zone composed of concentric diffractive rings. This combination of
zones creates two different focal points for light convergence, one for near objects and one for distant objects.
Hybrid IOLs are basically bifocal lenses (Fig. 4). The usual
strategy has been to have a distance component with a near
component targeting the usual near distance. However,
multifocal hybrid lenses are possible.
IOLs AND ACCOMODATION
New research into accommodating intraocular lenses indicates that many patients who get these implants will enjoy
good distance and near vision (52). The first hint that
intraocular lens implants could accommodate came back
in 1986, when Thornton used A-scan biometry to report on
anterior movement of a three-piece loop IOL (53). He found
that this forward movement allowed some patients to have
good uncorrected distance and near vision simultaneously.
The mechanisms of presbyopia remain incompletely
understood. A review of the variety of such mechanisms
has been presented by Atchison (54). Accommodation in the
youthful, phakic human eye is accomplished by contraction
of the ciliary body and subsequent release in the resting
tension of the zonular fibers by which the crystalline lens is
suspended, resulting in increased lens curvature (5557).
The weight of current evidence seems to suggest that
although some loss of ciliary body action might contribute
to reduced accommodation (58), significant ciliary body
function persists into advanced maturity, and that loss
of lens and capsule elasticity in concert with changes in the
geometry of zonular attachments are probably most culpable in producing the distress of presbyopia (59). If so,
replacement of the crystalline lens with a lens that
responds to ciliary body contraction should restore accommodative function (60). Attempts to replace the crystalline
lens by refilling the capsular bag with appropriately
deformable gels have been made (59,61,62). McLeod et al.
aimed at designing an accommodating intraocular lens with
extended accommodative range that can be adapted to
current standard phacoemulsification and endocapsular
implantation technique (60). They concluded that a dual
optic foldable IOL design can increase the optical effect of a
given displacement and suggests improvements for accommodating intraocular lenses.
BIBLIOGRAPHY
1. Barraquer JI. Keratomileusis for the correction of myopia.
Arch Soc Amer Oftalmol Optom 1964;5:2748.
240
LENSES, INTRAOCULAR
2. Ascher K. Prothetophakia two hundred years ago. Am J

Ophthalmol 1965;59:445446.
3. Nordlohne ME. The Intraocular Lens Development and
Results with Special Reference to the Binkhorst Lens, 2nd
ed. The Hague: Dr. W. Junk b.v.; 1975: p 1417
4. Ridley H. Intraocular acrylic lensespast, present and future.
Trans Ophthal Soc UK 1964;84:514.
5. Kwitko ML, Kelman CD, editors. Intraocular lens implantation: The Beginnings: The History of Modern Cataract Surgery. The Hague: Kugler Publications; 1998: p 3552.
6. Apple DJ, Sims J. Harold Ridley and the invention of the
intraocular lens. Surv Ophthalmol 1996;40:279292.
7. Rosen E. History in the making. J Cataract Refract Surg
1997;23:45.
8. Choyce P. Intraocular Lenses and Implants. London; H K
Lewis & Co. Ltd; 1964.
9. Apple DJ et al. Complications of intraocular lenses: A historical
and histopathological review. Surv Ophthalmol 1984;29: 154.
10. Javitt JC, et al. Outcomes of cataract extraction with multifocal intraocular lens implantation. Functional status and
quality of life. Ophthalmology 1997;104:589599.
11. Strampeli B. Sopportabilita` di lenti acrilichi in camera anteriore nella afachia e nei vizi di refrazione. Ann Oftalmol Clin
Ocul 1954;80:7582.
12. Barraquer J. Anterior chamber plastic lenses. Results and
conclusions from five years experience. Trans Ophthalmol Soc
UK 1959;79:393424.
13. Neuhann T, et al. Phakic intraocular lenses, J Refract Surg
1998;14:272279.
14. Baikoff G, Joly P. Correction chirurgicale de la myopie forte
par un implant de chambre anterior dans loeil phake. Bull Soc
Belge Ophthalmol 1989;233:109125.
15. Baikoff G. Phakic anterior chamber intraocular lenses. Int
Ophthalmol Clin 1991;31:7586.
16. Rosen ES, Haining WM, Arnott EJ editors. Intraocular Lens
Implantation. London, New York: Mosby; 1984.
17. Fyodorov SN, Durnev VV. Operation of dosaged dissection of
corneal circular ligament in cases of myopia of mild degree.
Ann Ophthalmol 1979;11:18851890.
18. Fyodorov SN, Zuev VK, Aznavayez VM. Intraokuliarnaia
korrektsia miopii vysokoi stepeni zadnekamernmi otritsatelimi linzami. Oftalmochirugia 1991;3:5758.
19. Fechner PU, Alpar J. Iris Claw Lens or Lobster Claw Lens of
Worst; 1986.
20. Fechner P, Worst J. A New concave intraocular lens for the
correction of myopia. Eur J Implant Refract Surg 1989;1:4143.
21. Hoerauf H, Menz DH, Dresp J, Laqua H. Use of 044 as a
solvent for silicone oil adhesions on intraocular lenses. J
Cataract Refract Surg 1999;25:13921396.
22. Ossipov A. Comparison of internal reflectance patterns of Collamer, acrylic and silicone. 1997. Data on file, STAAR Surgical.
23. Davis EA. Study of post-cataract surgery inflammation with 3
different IOLs (Collamer, SI40NB, AR40). Summary of data
found in all patients. Presented at OSN Meeting: New York:
October 2003.
24. Jia S, Wang X, Wang E. A study of suitable age for intraocular lens
implantation in children according to ocular anatomy and development, Zhonghua Yan Ke Za Zhi. 1996 Sept; 32(5): 336338.
25. Zou Y, Chen M, Lin Z, Yang W, Li S. Effect of cataract surgery
on ocular axial length elongation in young children. Yan Ke
Xue Bao 1998;14(1) : 1720.
26. Flitcroft DI, Knight-Nanan D, Bowell R, Lanigan B, OKeefe M.
Intraocular lenses in children: changes in axial length, corneal
curvature, and refraction. Br J Ophthalmol 1999;83(3):265269.
27. Kora Y, et al. Eye growth after cataract extraction and intraocular lens implantation in children. Ophthalmic Surg 1993;24(7):
467475.
28. Pan Y, Tang P. Refraction shift after intraocular lens implantation in children. Zhonghua Yan Ke Za Zhi 2001;37(5):328331.
29. Welford W. Aberrations of Optical Systems. Bristol: Adam
Hilger; 1962.
30. Born M, Wolf E. The Diffraction Principles of Optics: Electromagnetic Theory of Propagation, Interference, and Diffraction
of Light, 6th ed. New York: Pergamon Press; 1989.
31. Hamam H. Aberrations and their impact on image quality.
Wavefront Analysis, Aberrometers & Corneal Topography,
Agarwals edition, 2003.
32. Castejon-Mochon JF, Lopez-Gil N, Benito A, Artal P. Ocular
wave-front aberration statistics in a normal young population.
Vision Res 2002;42(13):16111617.
33. Howland HC, Howland B. A subjective method for the measurement of monochromatic aberrations of the eye. J Opt Soc
Am 1977;67(11):15081518.
34. Porter J, Guirao A, Cox IG, Williams DR Monochromatic
aberrations of the human eye in a large population. J Opt
Soc Am A Opt Image Sci Vis 2001;18(8):17931803.
35. Paquin MP, Hamam H, Simonet P, Objective measurement of
the optical aberrations for myopic eyes, Opt Vis Sci 2002;79:
285291.
36. Thibos LN, Hong X, Bradley A, Cheng X. Statistical variation
of aberration structure and image quality in a normal population of healthy eyes. J Opt Soc Am A 2002;19:23292348.
37. Francon M Vision dans un instrument entache daberration
spherique. The`se, editions de la Revue dOptique, 1945.
38. Smith WJ Modern optical engineering, the design of optical
system. New York: Me Graw-Hill, 1990.
tude des effets combines de la diffraction et des
39. Marechal A. E
aberrations geometriques sur limage dun point lumineux.
Revue dOptique; 1947.
40. Hamam H, New metric for optical performance. Opt Vis Sci
2003;80:175184.
41. Marsack JD, Thibos LN, Applegate RA Metrics of optical
quality derived from wave aberrations predict visual performance. J Vis 2004;4(4):322328.
42. Tognetto D, et al. Analysis of the optical quality of intraocular
lenses. Inv. Ophthalmol & Vis Sci (IOVS) 2004;45/8:26822690.
43. Negishi K, Ohnuma K, Hirayama N, Noda T. Effect of chromatic aberration on contrast sensitivity in pseudophakic eyes.
Arch Ophthalmol 2001;119:11541158.
44. Matin RG. Higher-Order Aberrations and Symptoms with
Pseudophakia, Symposium on Cataract, IOL and Refractive
Surgery, April 1216, 2003 San Francisco, CA.
45. Brunette I, et al. Optical quality of the eye with the artisan
phakic lens for the correction of high myopia. Optomol Vis Sci
2003 Feb; 80(2):167174.
46. Liang J, Grimm B, Goelz S, Bille JF. Objective measurement of
wave aberrations of the human eye with the use of a Hartmann-Shack wave-front sensor. JOSA A 1994;11:19491957.
47. Hamam H. An apparatus for the objective measurement of
ocular image quality in clinical conditions. Opt Commun
2000;173:2336.
48. Hamam H, de Bougrenet JL. Efficiency of programmable
quantized diffractive phase elements. Pure and Appl Opt
1996;5:389403.
49. Wilson K. New Technology Removes Cataract and Improves
Vision. Geriatr and Aging 1998;1: p 15.
50. Pieh S, et al. Quantitative performance of bifocal and multifocal intraocular lenses in a model eye: Point spread function in
multifocal intraocular lenses. Arch-Ophthalmol. 2002;120(1):
2328.
51. Malacara D, Malacara Z. Handbook of Lens Design. New York;
Marcel Dekker 1994.
52. Karpecki PM. The future of IOLs that accommodate. Rev Opt
Dec 2002.

53. Thornton S. Lens implantation with restored accommodation.
Curr Cana Ophthal Prac 1986;4:6062.
54. Atchison DA. Accommodation and presbyopia. Ophthal Physi
Opt 1995;15:255272.
55. Fisher RF. Presbyopia and the changes with age in the human
crystalline lens. J Physiol (London) 1973;228:765779.
56. Koretz JF, Handelman GH. Modeling age-related accommodative loss in the human eye. Math Mod 1986;7:10031014.
57. Schachar RA. Zonular function: A new model with clinical
implications. Ann Ophthalmol 1994;26:3638.
58. Hara T, et al. Accommodative intraocular lens with spring
action part 1. Design and placement in an excised animal eye.
Ophthal Surg 1990;21:128133.
59. Gilmartin B. The aetiology of presbyopia: A summary of the
role of lenticular and extralenticular structures. Ophthal
Physiol Opt 1995;15:431437.
60. McLeod SD, Portney V, Ting A. A dual optic accommodating
foldable intraocular lens. British J Ophthal 2003;87:10831085.
61. Cumming JS, Slade SG, Chayet A. AT-45 Study Group. Clinical evaluation of the model AT-45 silicone accommodating
intraocular lens. Results of feasibility and the initial phase of a
Food and Drug Administration clinical trial. Ophthalmology
2001;108:20052009.
62. Kuchle M, et al. Implantation of a new accommodating posterior chamber intraocular lens. J Refract Surg 2002;18:208216.
See also BIOMATERIALS: POLYMERS; CONTACT LENSES; VISUAL PROSTHESES.
241
that may be combined to produce mechanically stable

materials. The constituents are considered in terms of
the matrix in which are embedded fibers of collagen and
varying amounts of elastin. Following this is a discussion of
the ways these components interact in ligaments and
tendons to yield composite materials with the required
mechanical properties. A consideration of some of the ways
in which ligaments and tendons may be damaged, and the
mechanisms by which they might recover or be repaired,
leads to a final, brief review of their surgical replacement. A
small section on work being conducted in order to produce a
tissue engineered ligament and tendon is also included.
COMPONENTS
Tendons and ligaments are composed primarily of collagen
fibers surrounded by a matrix. Here the matrix refers to all
the materials that surround the collagen fibers providing
both structural support and a medium for diffusion of
nutrients and gases. Note this is in contrast to its use in
biological terms in which it generally includes the fibrous
components. The matrix contains proteoglycans and adhesive glycoproteins and is highly hydrated (typically 6580%
water) (5,6).
Collagen
LIFE SUPPORT. See CARDIOPULMONARY RESUSCITATION.

G AZANGWE
RM ASPDEN
INTRODUCTION
Tendons and ligaments are fibrous connective tissues that
play a mechanical role in the stability and locomotion of
the body by transmitting tension. Unlike muscle, which
actively contracts, ligaments and tendons are passive.
Tendons transmit mechanical forces from muscle to bone,
whereas ligaments join bone to bone. Both tendons and
ligaments contain relatively few cells (1), and their extracellular matrices are made up of several components.
These components are combined in various proportions,
and with different organizations to give mechanical properties appropriate to the function of the particular tendon
or ligament. There have been a number of reviews in recent
years covering specific ligaments, for example, in the rabbit
(2), or tendons, such as the human achilles (3), or aspects of
their behavior such as healing and repair (4). In this
article, the emphasis is on properties the ligaments have
in common, which will provide an insight into how and why
they behave as they do. This will be based around their
functioning as fiber-reinforced materials whose properties
are regulated by the cells they contain that produce and
maintain the extracellular matrix.
First, this article considers the components of the tissue,
not from a biochemical point of view, but as components
Collagen fibrils are able to reinforce the weak matrix

because of their much greater stiffness and strength in
tension (7). The collagen molecule is a long, stiff rod made
up of three polypeptide chains wound as a triple helix
structure (8). Each fibril is like a rope in which linear
molecules are packed together with their axes parallel,
within a few degrees, to the fibril axis. Molecules are held
together by covalent cross-links so that the fibril is very
strong in tension. The regular arrangement of molecules
along the axial direction in a fibril gives rise to the characteristic periodicity of 67 nm, which may be seen in
electron micrographs (Fig. 1). Although to date, 21 genetically different types of collagen have been identified, types
Figure 1. Electron micrograph of collagen fibril from rat tail tendon

stained with uranyl formate showing characteristic banding pattern
repeated every 67 nm. Micrograph courtesy of Dr D. P. Knight.
242
I and III fibrous collagens dominate in tendons and ligaments.

There is a heirarchical structure within tendons (9) that
has been reviewed often [e.g., (10)], and that appears to
have been based on an earlier description of keratin. In this
model, collagen molecules are arranged successively into
microfibrils, subfibrils, fibrils, and fascicles that are finally
packed into the tendon sheath. Evidence for micro- and
subfibrils is still equivocal, but a principal mechanical
structure is the fibril. These generally have a unimodal
distribution of diameters in the newborn, typically a few
tens of nanometers, but assume a bimodal distribution in
mature tendons and ligaments with mode diameters typically 100300 nm (11,12). The ends of fibrils are rarely
seen in electron micrographs, and even when they are seen,
it is not clear whether they are artifacts of sectioning.
Fibrils appear to grow by end-to-end addition of short
fibrils and there is evidence from 12-week old mouse skin
that fibril tips may eventually fuse with the central region
of other fibrils to create a meshwork (13). It is not known
whether this happens in tendon or ligament or whether the
fibrils are as long as the tendon or ligament. Fibrils are
arranged into fascicles, which are 80300 mm in diameter
and these have a crimped morphology that is seen most
clearly using polarized light (6,9). This crimp is generally
described as a planar zigzag with a sharp change in direction of the collagen fibrils with a periodicity of 200
300 mm (Fig. 2). On application of a tensile load, initial
elongation of the fiber requires a relatively low stress
because it simply leads to removal of the crimp (14,15).
Once the crimp is removed, the fibers become very stiff.
This crimp structure, which is also found in ligaments,
explains the characteristic shape of the stressstrain curve
for ligaments and tendons (14).
There are many studies of the mechanical properties of
tendon and ligament. Most tendons have similar properties, because of their role transmitting forces from muscle
to bone and the need to do this most efficiently (1619). In
contrast, every ligament has unique properties that fit it to
its function at that particular site in the body. These may
range, for example, from the highly extensible ligamentum
flavum, helping control spinal posture, to the relatively
stiff cruciate ligaments in the knee. Most information on
the mechanical properties of collagen has been inferred
from experiments on tendon; and though they contain a
large proportion of collagen, 7080% of the dry weight,
the fibrils are surrounded by matrix, and therefore
(a)
(b)
the tissue is really a composite material (6). This makes

it difficult to separate the behavior of the individual
components.
Proteoglycans
The proteoglycans found predominantly in tendon and
ligament belong to the small leucine-rich proteoglycan
(SLRP) family; decorin, biglycan, lumican, and fibromodulin though there are small amounts of the large proteoglycans aggrecan (20,21) and versican (21,22). The SLRPs all
comprise a repeating structure that is rich in leucine
residues, between 13 and 16% of all residues (23). They
are present in different tissues in differing amounts and
appear at different stages of development (24). Their function is poorly understood though gene knockout studies in
mice have shown marked osteopenic effects on bone, skin
laxity, and irregular collagen fibers in tendon (25). Most of
these SLRPs have one or two glycosaminoglycan (GAG)
chains of varying lengths attached, generally either dermatan sulfate or chondroitin sulfate, which can interact
electrostatically with collagen (26,27). The proteoglycan
decorin has been found to be localized in the tissue to a
specific region of a collagen fibril (28). It is also reported to
play a regulatory role in collagen fibrillogenesis, by affecting fibril radius (13), and increases the strength of uncrosslinked fibers (29,30). In regions where tendons pass over
bone and are subjected to compressive loading in addition
to tension, a fibrocartilaginous tissue containing large
amounts of aggrecan and biglycan develops (31,32). The
adhesive glycoproteins include fibronectin and thrombospondin (3335), both of which contain possible attachment
sites for cells. In humans, these were reported to be more
common in the tendon sheath than in the fibrous bulk (36)
and fibronectin was found to be up-regulated at sites of
injury (34).
The matrix is weak in shear; that is, if it is loaded in
compression, it tries to slide out sideways unless it is
contained. This behavior may be readily seen in jelly (Jello
in the United States) and is not surprising given its high
water content, since fluids simply flow when sheared. It is
not easy to measure the shear strength of matrix. Proteoglycans at similar concentrations have a very low shear
strength (37); however, matrix may be stiffer than this
because of the interactions between its macromolecular
components. An analysis of the behavior of tendon suggests
that its matrix would have a shear modulus of 100 kPa
(38). Because of this low stiffness in shear, the matrix alone
is not well suited to bearing loads. Also, its proportion in
ligament and tendon is quite low, 1020% of the dry
weight. The ways in which matrix may transmit stress
to the fibers and prevent crack propagation will be discussed later.
(c)
Elastic Fibers
Figure 2. Schematic diagram of the crimp structure in collagen

fibers seen from the side. (a) relaxed state [this is grossly
exaggerated; the true crimp angle (u) is 158]. As the fiber is
stretched, the crimp straightens out (b) until at strains of a
0.03, the fiber becomes straight (c), at which point it becomes
much stiffer.
Electron microscopy reveals the presence of elastic fibers in

ligaments and tendons (39,40). Elastic fibers have two
components; elastic fiber microfibrils and elastin. The
microfibrils have a diameter of 1012 nm and consist of
glycoproteins. Elastin is composed of mainly hydrophobic
nonpolar amino acids with a high content of valine (41).
Elastic fibers are highly extensible, they do not creep and

their extension is reversible at high strains. Their mechanical properties are thus very different from collagen. Most
of our knowledge of elastic fibers comes from experiments
on ligamentum nuchae, a ligament from the cervical spine,
which contains 70% elastin by dry weight (42). Elastin
closely resembles a rubber in many respects and its
mechanical properties are certainly very similar (43). Purified samples of ligamentum nuchae will extend to roughly
twice their resting length before breaking.
The extensibility of a tendon or ligament depend in part
on the elastin content of the tissue. Ligamentum flavum,
from the spine, which may typically be strained to 50%
contains roughly 70% elastin, by dry weight (44), whereas
tendon, which works at strains <4% contains only 2%
elastic fibers by dry weight (1). It is fairly easy to see
why highly extensible tissues have a high proportion of
elastin, but not quite as easy to explain the presence of
elastic fibers in a relatively inextensible tissue such as
tendon. A clue is provided by some synthetic fibrous composite materials that contain two different kinds of fiber
(45). Here a small proportion of strong, low stiffness fibers
added to the composite produces a material that is less
susceptible to failure under sudden application of load than
one that contains only stiff fibers; that is, it makes the
material less brittle. It may be that the small proportion of
elastic fibers in tendon provide some protection against the
sudden application of load that may occur, for example, if
an animal is startled.
FiberMatrix Interactions
The combination of strong fibers in a weak matrix leads to
materials that are less susceptible to mechanical damage
while maintaining a high proportion of the strength of the
fibers. In particular, they are less susceptible to sudden
failure than a homogeneous material would be; a property
called toughness (46). This composite nature has been
recognized for many years (6) and provides a theoretical
framework for understanding the properties of the tissues.
It also enables some useful comparisons to be made
between relatively simple synthetic composites and biological tissues in which the complexities of composition and
structure make modeling very difficult. The aim is to obtain
an understanding of how the similarities in the tissues,
fibers in a matrix, enable them to function in general terms
before considering the differences (in composition, and
organization), which give them their specific properties.
The function of collagen fibrils and fibers in such a composite is to withstand axial tension, since, like any rope, they
have little resistance compression and flexion (7). As the
tissue is stretched the matrix will try to flow and this will
exert a shear force along the surface of the collagen fibers
tending to stretch and orient them (7,47). This length
increase, which is normally expressed as a fraction of
the original length and is then termed strain, leads to
a restoring force in the fiber that balances the applied force.
The behavior is rather like a loaded spring that stretches to
enable it to bear load, but returns to its relaxed length on
removing the load. Similarly, collagen fibers are able to
reinforce a tissue if they are oriented so that an applied
243
load tends to stretch them. The nature of the shear force

exerted by the gel is unknown, but two simple models,
those of elastic and plastic (or frictional) stress transfer,
have been used to investigate stresses in the fibers and the
force that has to be generated at the fiber surface to enable
them to function in this way (4749). Fibers that are
shorter than the tissue are still able to provide reinforcement (50). Some fibrils observed in tissues (51) and those
grown in vitro (52,53) appear to be tapered, rather than
having a uniform radius. Analytical and finite element
models of idealized single-fiber composites have shown
that tapered fibers have two distinct advantages over uniform fibers: the axial stresses within the fiber are more
uniformly distributed and they contain a much smaller
amount of material, though their effectiveness at reinforcing is just as great. A more uniform stress within the fiber
means more of the fiber is carrying a significant stress, thus
making better use of the fiber, and avoids the generation of
stress concentrations that are potentially damaging and
could lead to fiber fracture. In addition, the volume of
material in a cone, for example, is only one-third of that
in a straight cylinder and, therefore, a tapered fiber incurs
a far smaller metabolic cost by the cells to produce it. In
straight-cylindrical fibers it has been calculated that interactions at the fiber surface do not have to be great in order
to load fully the fiber. In tendon, assuming conservatively
only one interaction per 67 nm D-period, it was estimated
that fibermatrix interaction forces of the order of only 10
pN was sufficient to load fully the fiber (47). These forces
are similar in magnitude to van der Waals forces or hydrogen bonds. This suggests that permanent bonds or covalent
interactions between fiber and matrix are not essential for
the mechanical functioning of the tissue though, of course,
it does not preclude them. Regulating the interaction
between fibers and matrix is clearly important in this
model of how the tissues function and decorin, as described
above, is a prime candidate for a role in this. Changes in the
concentration and orientation of collagen and its interactions with the matrix have been used to explain the dramatic changes in a similar fibrous tissue, the uterine
cervix, that occur during parturition (54). The presence
of the matrix around the collagen fibrils is also important
when it comes to preventing crack propagation. This will be
considered in more detail in the context of the tissues
themselves.
LIGAMENT
Ligaments are short bands of tough, but flexible, fibrous
connective tissue that bind bones together and guide joint
motion, or support organs in place. The word ligament is
derived from the Latin word ligare, which means to bind.
Generally, ligaments can be classified into two major subgroups. There are those connecting the elements of the
skeletal system (usually crossing joints) and those connecting other soft tissues, such as the suspensory ligaments in
the abdomen. This section only considers skeletal ligaments. The main function of the skeletal ligaments, such
as the anterior cruciate ligament (ACL) of the knee joint, is
to stabilize and control normal kinematics, to prevent
244
Figure 3. (a) Video image showing a ligament

failing under tensile strain. (b) A schematic diagram
showing a longitudinal section through a fiber composite illustrating how a weak matrix prevents crack
propagation from one fiber to another and dissipates its
energy by creating cracks in directions other than
across the composite.
whereas the ligamentum flavum of the human spine operates at strains of up to 0.6. Figure 4 shows that ligamentum
flavum is much less stiff than tendon.
It is not surprising that some ligaments contain a high
proportion of elastin (6070% of the dry weight), which
enables them to withstand the high strains to which they
are subjected without fracture (61). Ligaments are viscoelastic, that is their properties are time dependent and they
appear stiffer if stretched more rapidly. These ligaments
exhibit hysteresis, that is, they lose energy on being taken
through a cycle of stretching and relaxing. Tkaczuk (61)
published a detailed account of the mechanical properties
of longitudinal ligaments from the human spine. These
deform elastically up to strains of 0.25, when the stress is
5 MPa, and rupture at a stress of 20 MPa. Shah
et al.(62) also showed that, like tendons, the collagen fibers
are crimped and this crimp disappears at strains of 1.2
2.8% depending on the ligament. When ligaments are cut
from the joint, they can often be seen to contract rapidly,
suggesting that they are held in a state of tension even
when the joint is in a relaxed state.
100
80
Tendon
Force (N)
abnormal displacements and rotation that may damage the

articular surfaces (2,55). At the insertion to bone, the
ligaments change from flexible ligamentous tissue to rigid
bone, mediated by a transitional zone of fibrocartilage and
mineralized cartilage. This helps to prevent stress concentration at the attachment site by allowing gradual change
in stiffness (56,57).
Similar to tendon, ligaments are primarily composed of
collagen embedded in a weak matrix (7). The collagen
molecules in ligaments pack together to form fibrils and
the fibrils aggregate into larger fibrous bundles (2). As
described previously, the function of collagen fibrils is to
provide tensile reinforcing for the matrix. The proportion of
fibers within the ligamentous structure and the orientation
of the collagen fibers are the main factors that govern the
mechanical behavior of the tissue (7). In contrast to tendon,
there is commonly less collagen, which is less highly
oriented than in tendon, conferring a generally greater
extensibility to these tissues.
The collagen fibrils also prevent damaged tissues from
failing suddenly. For example, most ligaments do not tear
straight across when they are damaged (Fig. 3). Instead,
small tears in the matrix are diverted when they encounter
the strong collagen fibrils (see below on failure). There is
then a possibility that a damaged ligament can heal while,
in the meantime, retaining the ability to withstand some
load. Some ligaments, however, (e.g., the ACL), have a
limited ability to heal when ruptured and this means
that it often needs to be replaced or reconstructed when
ruptured. A brief summary of different options available
for treating ruptured ligaments will be presented in a
later section.
Unlike tendons that all have very similar composition,
structure, and function, the same cannot be said of ligaments, and it is far harder to make general comments
about their properties. Much less is known, too, about
the relationship between their structures and functions
as the arrangements of their collagen fibrils are more
complex than those in tendons (58,59). This greater complexity is understandable when it is realized that the
function of a ligament is very dependent on its position
in the body; for example, the medial collateral ligament in
the knee of a sheep operates at strains of 0.02 (60),
60
40
Ligamentum
flavum
20
0
0
0.2
0.4
0.6
0.8
1.0
Strain
Figure 4. Comparison of forcestrain curves obtained for
extension of tendon and ligamentum flavum.
and are viscoelastic. The collagen fibers are relatively

disoriented in the unstretched tissue and become more
highly aligned as the tissue is stretched. They often contain
a proportion of elastin. Their composition and structure
depend on their position in the body and their dynamic
behavior, that is, the change in structure with strain,
becomes more important.
100
TENDON
60
40
20
0.02
0.04
0.06
Strain
Figure 5. Full width at half-maximum (fwhm), D, of the distribution of orientations of collagen fibers in ligamentum flavum. (o)
and posterior longitudinal ligament, () as a function of strain.
Spinal ligaments provide a good example of the mechanical function of ligaments in a joint (58). The longitudinal
ligaments and the ligamentum flavum act together with
the intervertebral disc to achieve a mechanically stable
joint and serve to limit its mobility. The ligamentum
flavum is almost twice as far from the axis of rotation in
forward bending as the posterior longitudinal ligament,
and hence it can be seen that it needs to be roughly twice as
extensible (58,63). This is partly explained by a higher
elastin and lower collagen content in ligamentum flavum
(61), but also by a less highly aligned organization of
collagen fibers (63). As the ligament is stretched, the fibers
become more highly aligned and this will increase the
stiffness of the tissue, that is its resistance to extension.
X-ray diffraction experiments have measured the spread of
orientations of fibrils in these ligaments and modelling has
shown how this decreases with increasing extension (47).
Figure 5 shows that the width of the distribution, Du, as
measured at half the peak height, is greater for ligamentum flavum than posterior longitudinal ligaments. This
mechanism provides an explanation for the form of the
forcestrain curve, shown in Fig. 4, One final point about
ligaments is that they have a nerve supply that makes
them potential sources of pain. It has also been suggested
that they may function as proprioceptors as part of a reflex
arc, that is, the ligaments would act as sensors to detect the
position of a joint and the information would then be used
to control the muscles around the joint thereby controlling
its movement and stability (64).
In summary, ligaments are composite materials containing crimped collagen fibers that are prestressed in the
relaxed joint. They have a nonlinear stressstrain curve
The function of tendon is to transmit the force generated by

a contracting muscle to the correct point of application on a
bone so as to manipulate a joint. Tendons are often preferable to direct attachment of muscle to bone because of
various functional requirements. Muscles have a low tensile strength, defined as load at fracture per unit crosssectional area. This means that they must have a large
cross-sectional area in order to transmit sufficient force
without tearing. Around many joints (e.g., the fingers),
there is insufficient space to attach many muscles. The
muscle, therefore, is located further away and attachment
made by a tendon, which may be tens of centimeters long in
the hand and forearm. Tendons, therefore, need to be
strong, so that they can be relatively slender, and stiff,
so that the force developed by the muscle is transmitted to
the bone without energy being wasted on stretching the
tendon. A graph of force as a function of strain for a tendon
is shown in Fig. 6. This type of curve is obtained by
subjecting the tendon to various forces and recording the
amount by which the tendon is stretched, and many of
the early studies have never been bettered (6,9,18,40,65).
The steepness of the stressstrain curve is a measure of the
stiffness of the material. Figure 6 shows that for small
strains the tendon requires very little force to stretch it but
thereafter it stiffens considerably. The stress at this point
is 10 MPa, which is several orders of magnitude greater
than the stresses needed to shear the matrix (6). If the
stiffness did not increase, a muscle would continue to
stretch the tendon and the force would not be transmitted
to the bone.
Tendons are believed to function in the body at strains
up to 0.04 (66,67). Beyond this strain, a tendon does not
return to its original length when the applied stress is
removed. A tendon will break at strains of 0.1 (67). The
30
20
Force (N)
degrees
80
245
10
0
0
0.02
0.04
Strain
0.06
Figure 6. Forcestrain curve for tendon.
stress
Stress
246
Strain
Strain
Figure 7. Schematic of stressstrain showing energy stored in a

stretched tendon.
Stress
initial stages of tendon extension involve straightening the

crimp of the collagen fibers described previously.
The energy stored in the stretched tendon is given by the
area under the stressstrain curve, as shown in Fig. 7, and
it is this energy that must be used to create a tear in the
tissue. This energy is lower in a material with a J-shaped
stressstrain relationship than if, for example, it were
linear. Minimizing the energy available to cause a fracture
in this way gives the tissue a property known as resilience,
that is, a tendon does not suddenly fail if it is overloaded
unlike a steel wire. While the crimped collagen fibers are
being straightened, the weak matrix must be sheared.
Because of the fluid-like nature of the matrix, it tends to
flow. The rate of flow depends on the force applied to it and
the amount of flow is greater the longer the force is applied.
The result is that tendons are viscoelastic (18,38,68). The
effect of the rate at which force is applied to the tendon is
shown in Fig. 8.
This time dependence of mechanical behavior leads to a
phenomenon known as creep. For example, when a load
of 10 N was applied to a human flexor digitorum tendon,
the initial strain was 0.015, but 100 s later under the same
load, the strain had increased to 0.016 (69). Thus, if a
0.02
0.04
0.06
Strain
Figure 8. Schematic stressstrain curves for tendon to show the
effects of different loading rates. These curves correspond to slow
loading (continuous curve) and rapid loading (dashed curve).
Figure 9. Stressstrain curves for tendon that is stretched

(continuous curve) and then relaxed (dashed curve), showing
hysteresis; that is, not as much energy is recovered on relaxing
as was initially used to stretch the tissue.
tendon is stretched rapidly, the matrix has less chance

to flow and creep in the material is less resulting in the
tendon being stiffer as shown in Fig. 8. Viscous flow of the
matrix also provides a mechanism for dissipating energy;
more work is done in stretching a tendon than is recovered
when the tendon is allowed to relax. This phenomenon is
known as hysteresis and is illustrated in Fig. 9. The behavior of a tendon is therefore intermediate between that of
a steel wire that stores all the energy used to stretch it, and a
viscous liquid that simply flows to a new position and does
not store any of the energy put in to cause it to flow.
MEASURING THE PROPERTIES OF LIGAMENTS
AND TENDONS
To quantify the physical properties of ligaments and tendons, mechanical testing of boneligament/tendonbone
complexes is often performed (7073). That this method
is often used is partly due to the difficulty in testing
isolated ligaments and tendons. Ideally, testing isolated
ligaments and tendons would provide measures of the
material properties of the tissue alone, but such tests
are complicated by difficulties in effectively securing the
cut ends (2). Putting the free ends in clamps often results in
stress concentrations at the grips, which may contribute to
premature failure. Although the use of boneligament/
tendonbone complexes still provides more secure clamping, it also increases the difficulty of separating the properties of the ligament from those of the insertion sites. When
such complexes are subjected to tensile loading, the resulting load-displacement curve represents the mechanical
properties of the boneligament/tendonbone complex as
whole rather than specifically about the material that
makes up the ligament or tendon.
In order to obtain the material properties of the ligaments, one needs to measure their length (to calculate
strain from deformation divided by original length) and
cross-sectional area (to calculate stress from applied force
divided by original area). From stressstrain curves material properties such as Youngs modulus (slope of stress
strain curve), maximum stress, maximum strain, and
energy density (area under stressstrain curve) can be
determined. Special devices such as buckle transducers

(74) and Hall-effect displacement transducers (75) have
been used to measure ligament strains during testing. The
drawback of such devices is that they rely on direct contact
with the tissue sample and that may influence the results.
Optical analysers have also been employed as a noncontact
method to measure ligament strains (2). However, inaccuracies may occur because the irregular dye blobs used as
markers change shape on stretching making it difficult to
define unique points. Another technique that has been
employed to measure strain is the use of video dimension
analyser (VDA) (76,77). This method requires no direct
contact with the specimen, but relies on a recorded video
image.
The irregular and complex shape and geometry of ligaments and tendons also make it difficult to measure their
cross-sectional area. Although flexible callipers, which are
able to follow contours better, have been used to measure
ligament cross-sectional areas (2), they still require contact, and this results in errors in measurements. Other
investigators have calculated the cross-sectional area of a
known length of ligament from measurements of its density by floatation in a mixture of xylene and carbon tetrachloride (78). A number of noncontact methods, such as the
use of a rotating microscope (79), use of the VDA (80) and
the laser micrometer (81) have also been employed to
measure the cross-sectional area.
When mechanical properties of ligaments and tendons
are being determined, it is important to consider the rate at
which they are loaded since they are viscoelastic, that is,
their mechanical properties depend on the rate at which
they are deformed (8286). This sensitivity to strain rate
means that ligaments and tendons exhibit properties of
stress relaxation (decreased stress with time under constant deformation) and creep (increased deformation with
time under constant load) (87,88).
Because of the complex geometry of tendons and ligaments, the orientation of the specimens during mechanical testing affects their physical properties and the
manner in which they fail, and should therefore be taken
into account when performing mechanical tests. Torsion
has been implicated as a factor in the rupture of the ACL
during sporting injuries (8991). Cyclic loading has also
been found to lower the yield point or soften the ligament
by increasing its compliance (decrease the slope of the
linear region of the stressstrain curve) (55). Azangwe
et al. (92) showed that, when combined, tensiontorsion
loading affects both structural and mechanical properties
of anterior cruciate ligaments.
LIGAMENT AND TENDON FAILURE MECHANISMS
Since ligaments and tendons consist of collagen fibers
reinforcing a weak matrix, it is reasonable to compare
their behavior under tensile loading with that of synthetic
fiber-reinforced composites, since their failure mechanisms are well established. This section describes some
of the modes of failure of fiber-reinforced composites
and how they are related to those of the ligaments and
tendons. Detailed accounts of failure modes of synthetic
247
fiber-reinforced composites can be found in textbooks, for

example, Agarwal and Broutman (45), and Kelly and
Macmillan (50). When a material is subjected to any kind
of loading, it can absorb energy by two basic mechanisms:
1. Material deformation.
2. Creation of new surfaces.
Material deformation occurs whenever a material is subjected to load. However, if the energy supplied is sufficiently large, cracks may be initiated. Whether they
propagate depends on the relative amounts of energy
required to create new surfaces compared with that stored
in the deformed matrix. For brittle materials such as glass,
the energy required to create a new fracture surface is
small and though only a small amount of deformation takes
place, this is elastic and the associated energy is sufficient
to propagate the crack. This means that brittle materials
have a low energy-absorption capability. On the other
hand, for ductile materials, large plastic deformations
occur, which dissipate energy, resulting in large energies
being absorbed rather than being available to drive a
fracture. This finding shows that the total energy-absorbing capability or toughness of a material can be enhanced
by increasing either the length of the path of the crack
during separation or the material-deformation capability.
In metals, the latter mechanism frequently occurs and
metallurgical processes are developed to maintain ductility. In composites, replacing low energy-absorbing constituents with greater energy-absorbing constituents can
enhance the toughness.
As is the case with many materials, failure in a fiberreinforced composite emanates from small inherent defects
in the material. Several failure events may occur during
the fracture of a fiber reinforced composite material,
such as,
1. Microcracking of the matrix.
2. Separation of fibers from the matrix (debonding and
pull-out).
3. Breaking of fibers,
Forcing a crack to take a longer path is the main mechanism
encountered in fiber composites to increase toughness. In
fibrous materials, when a matrix crack encounters a strong
fiber placed perpendicular to the direction of crack propagation, if the crack cannot cross the fiber because it is too
strong, the crack is forced to branch to run parallel with the
fiber (Fig. 10). If the crack goes right around the fiber, this
may result in fiber debonding, and pull-out, that is, becoming detached from the matrix and pulled out leaving fiber
ends showing as the material is stretched. In many cases
the surface area produced by secondary cracks is much
larger than the area of the primary cracks. This may
increase the fracture energy many times and is an effective
way of increasing the toughness of composites or the total
energy absorbed during fracture. Fibers may eventually
fracture when their strength is exceeded. For most
synthetic fiber-reinforced composites, fibers are separated
by matrix and therefore are unable to pass energy directly
248
this is that there are differences in healing characteristics

between different ligaments. The ACL of the knee joint, for
example, appears to have a poor healing capacity when
injured prompting the need for reconstruction. At the
junctions of ligaments to bone, the fibers of the ligament
become more compact, then cartilaginous and finally calcified before finally merging into the bone and there are
changes in cell phenotype and expressed proteins (96,97).
This complex structure reflects the difficulty of attaching a
tough, flexible material to a hard, brittle one.
LIGAMENT AND TENDON REPAIR
Figure 10. Model of crack tip in a fiber-reinforced composite

showing how a crack may propagate Fibers become debonded
from the matrix and are pulled out as the crack widens.
from one to another, hence it is unlikely that they all fail

together. Collagen fibers in ligaments are similarly separated by the matrix, and hence the transmission of stress
from fiber to fiber in the ligament is indirect (7). The
situation is more complicated in biological tissues because
of the complexity of the components and their arrangement
within the tissue and because of the possibility of repair. It
seems reasonable to assume that a crack will start first in
the weak matrix rather than in the strong collagen fibers.
It is unlikely that the crack will spread into a fibril, as
previously discussed, but it will be deflected by the fibrils
into new directions.
It appears that tendons and ligaments can continue to
withstand stress long after they are damaged, but before
complete fracture occurs (70,71,93). However, since
damage implies an irreversible change, at least until the
biological repair process begins, the tissue will not return
to its original dimension when the stress is removed. When
testing bonetendon/ligamentbone complexes, an additional failure mode may occur at the ligament insertion
to bone. For example, the ligamentum flavum tears at the
enthesis, the junction with the bone, leaving virtually no
possibility of natural healing (72). Fortunately, this injury
does not appear to occur in vivo and can occur in isolation or
in combination with other failure modes. As mentioned
previously, most ligaments appear to be prestressed within
the body so that if they are severed or avulsed they retract
and the damaged ends are no longer in contact. This makes
it very difficult for cells within the tissue then to bridge the
gap by synthesizing new matrix.
The tearing of tendons is a fairly common injury, though
rupture of the tendon tends to occur at the junction with
bone. Healing of torn tendons in humans is slow, and is not
always improved by surgical intervention (3). Healing
starts with the invasion of cells into the damaged area
that first lay down fine, poorly oriented collagen fibrils (94).
These fibrils later increase in diameter and become increasingly oriented as the cell population drops and the structure
becomes more akin to that of the original tendon (95).
Damage to and repair of ligaments follows a very similar
pattern to that described for tendons. A point to note in all
Because of the slow rate of healing of certain ligaments and

tendons, their reconstruction with synthetic materials has
been attempted, with varying degrees of success, for many
years (98). The area of prosthetic materials and methods is
so vast that only a very brief survey will be attempted here.
The main approaches that have been tried are replacement
using tissues taken from another part of the body, or from
an animal, and complete substitution by a synthetic material. None of these approaches has so far proved entirely
satisfactory. Success rates of 8090% are reported for both
techniques (99101), but some long-term studies have
shown that this success rate may fall to 4050% after
5 years (102) and there are reports of high, 50%, incidence of degenerative change (101,103). Friedman et al.
(104), reviewed the autogenous reconstruction of the ACL
using patellar tendon, iliotibial band, gracilis tendon, semitendinosus, or meniscus that gives some idea of the range
of tissues that have been tried. Unfortunately, most repairs
stretch with time resulting in loss of stability (105); this is
probably due to the differences in structure and mechanical properties between the original tissue and the replacement, described earlier. If the replacement tissue is
stretched too much while being fixed in position, it may
be irreversibly strained.
Early attempts at the synthetic replacement of ligaments and tendons, using silk, for example, were not
successful. These prostheses were intended to be permanent, but had not the strength and fatigue resistance to
withstand the millions of cycle of loading imposed on them
during the lifetime of the recipient. More recently polyester
(106), carbon fiber (107,108), or various combinations of
synthetic materials and autogenous tissue have all been
tried but still seem not to overcome this particular problem
(103,109,110).
Another approach that is currently being explored for
augmenting or reproducing ligaments and tendons is that
of tissue engineering (111,113), though this has not yet
reached the stage of clinical utility. These techniques may
include the development of biodegradable scaffolds, on which
it is hoped to encourage cells from the patient to grow a
replacement tissue (113), and growth factors (114) and their
introduction into the tissue using gene transfer (112,115).
SUMMARY
Tendons and ligament are connective tissues subject primarily to tensile forces. They comprise crimped collagen
fibrils and some elastin embedded in a weak matrix. Collagen fibers in tendon, composed of bundles of fibrils, are
highly aligned along the direction of applied force. The
initial structural response to the application of force is
straightening of the crimp, which occurs at a strain of
2%, after which the tendon stiffens considerably. Higher
strains are not immediately reversibly and may lead to
structural damage to the tissue. The function of the crimp
appears to be to minimize the energy stored in the
stretched tissue thus reducing the energy available to
cause fracture. Many ligaments can be stretched more
than tendons, because of their more complex collagen fibril
organization and, sometimes, the high proportion of elastin
present. Tendons and ligaments are viscoelastic; they do
not store all the energy used to stretch them, their response
to load depends on the rate at which it is applied, and they
continue to deform even if the applied load remains constant. Because a lot of energy is required to produce a large
fracture surface, they do not break easily, that is, they are
tough materials. Because some ligaments do not heal well
when injured, there is a need to replace them. However,
success in this area is still limited. Future replacements
may include tissue engineered ligaments.
ACKNOWLEDGMENTS
We thank Professor D.W.L. Hukins for valuable discussions over many years and the many other colleagues who
have contributed to some of the work we have described
here. We also thank the Medical Research Council, the
Engineering and Physical Sciences Research Council, and
the Arthritis Research Campaign for financial support.
BIBLIOGRAPHY
1. Fank CB, Shrive NG. Ligament. In: Nigg BM, Herzog W,
editors. Biomechanics of the Musculo-Skeletal System. Chichester: J Wiley; 1999.
2. Frank CB, Hart DA, Shrive NG. Molecular biology and
biomechanics of normal and healing ligamentsa review.
Osteoarthritis Cartilage 1999;7:130140.
3. Maffulli N. Rupture of the Achilles tendon. J Bone Joint Surg
1999;81-A:10191036.
4. Woo SL, Debski RE, Zeminski J, Abramowitch SD, Saw SS,
Fenwick JA. Injury and repair of ligaments and tendons.
Annu Rev Biomed Eng 2000;2:83118.
5. Nachemson AL, Evans JH. Some mechanical properties of the
third human lumbar interlaminar ligament (ligamentum
flavum). J Biomech 1968;1:211220.
6. Elliott DH. Structure and function of mammalian tendon.
Biol Rev 1965;40:392421.
7. Hukins DWL, Aspden RM. Composition and properties of
connective tissues. Trends Biochem Sci 1985;10:260264.
8. Brodsky B, Ramshaw JA. The collagen triple-helix structure
(Review). Matrix Biol 1997;15:545554.
9. Kastelic J, Galeski A, Baer E. The multicomposite structure
of tendon. Connect Tissue Res 1978;6:1123.
10. Vincent JFV. Structural Biomaterials Princeton. 2nd ed.
New Jersey: Princeton University Press; 1990.
11. Parry DA, Barnes GR, Craig AS. A comparison of the size
distribution of collagen fibrils in connective tissues as a
function of age and a possible relation between fibril size
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
249
distribution and mechanical properties. Proc R Soc London

Ser B Biol Sci 1978;203:305321.
Patterson-Kane JC, Wilson AM, Firth EC, Parry DA, Goodship AE. Comparison of collagen fibril populations in the
superficial digital flexor tendons of exercised and nonexercised thoroughbreds. Equine Vet J 1997;29:121125.
Kadler KE, Holmes DF, Graham H, Starborg T. Tip-mediated
fusion involving unipolar collagen fibrils accounts for rapid
fibril elongation, the occurrence of fibrillar branched networks in skin and the paucity of collagen fibril ends in
vertebrates. Matrix Biol 2000;19:359365.
Buckley CP, Lloyd DW, Konopasek M. On the Deformation of
Slender Filaments with Planar Crimp: Theory, Numerical
Solution and Applications to Tendon Collagen and Textile
Materials. Proc R Soc London Ser A, Math Phys Sci 1980;
372:3364.
Diamant J, Keller A, Baer E, Litt M, Arridge RGC. Ultrastructure of collagen as a function of ageing. Proc R Soc
London Ser B Biol Sci 1972;180:293312.
Betsch DF, Baer E. Structure and mechanical properties of
rat tail tendon. Biorheology 1980;17:8394.
Haut RC, Lancaster RL, Decamp C. Mechanical properties of
the canine patellar tendonsome correlations with age and
the content of collagen. J Biomech 1992;25:163.
Woo SL. Mechanical properties of tendons and ligaments. I.
Quasi-static and nonlinear viscoelastic properties. Biorheology 1982;19:385396.
Woo SLY, Sites TJ. Current advances on the study of the
biomechanical properties of tendons and ligaments. In:
Nimni ME, editor. Collagen, Volume II, Biochemistry Biomechanics. Boca Raton: CRC Press; 1988.
Campbell MA, Tester AM, Handley CJ, Checkley GJ, Chow
GL, Cant AE, Winter AD, Cain WE. Characterization of a large
chondroitin sulfate proteoglycan present in bovine collateral
ligament. Arch Biochem Biophys 1996;329:181190.
Robbins JR, Vogel K. Regional expression of mRNA for proteoglycans and collagen in tendon. Eur J Cell Biol 1994;64:
264270.
Thomopoulos S, Hattersley G, Rosen V, Mertens M, Galatz L,
Williams GR, Soslowsky LJ. The localized expression of extracellular matrix components in healing tendon insertion sites:
an in situ hybridization study. J Orthop Res 2002;20:454463.
Hocking AM, Shinomura T, McQuillan DJ. Leucine-rich
repeat glycoproteins of the extracellular matrix. Matrix Biol
1998;17:119.
Wilda M, Bachner D, Just W, Geerkens C, Kraus P, Vogel W,
Hameister H. A comparison of the expression pattern of five
genes of the family of small leucine-rich proteoglycans during
mouse development. J Bone Miner Res 2000;15:21872196.
Corsi A, Xu T, Chen XD, Boyde A, Liang J, Mankani M,
Sommer B, Iozzo RV, Eichstetter I, Robey PG, Bianco P, Young
MF. Phenotypic effects of biglycan deficiency are linked to
collagen fibril abnormalities, are synergized by decorin deficiency, and mimic Ehlers-Danlos-like changes in bone and
other connective tissues. J Bone Miner Res 2002;17:11801189.
Gelman RA, Blackwell J. Collagen-mucopolysaccharide
interactions at acid pH. Biochim Biophys Acta 1974;342:
254261.
Lindahl U, Hook M. Glycosaminoglycans and their binding
to biological macromolecules. Annu Rev Biochem 1978;47:
385417.
Scott JE, Orford CR. Dermatan sulphate-rich proteoglycan
associates with rat tail tendon collagen at the d band in the
gap region. Biochem J 1981;197:213216.
Weber IT, Harrison RW, Iozzo RV. Model structure of decorin
and implications for collagen fibrillogenesis. J Biol Chem
1996;271:3176731770.
250
30. Pins GD, Christiansen DL, Patel R, Silver FH. Self-assembly

of collagen fibers. Influence of fibrillar alignment and decorin
on mechanical properties. Biophys J 1997;73:21642172.
31. Evanko SP, Vogel KG. Proteoglycan synthesis in fetal tendon
is differentially regulated by cyclic compression in vitro.
Arch Biochem Biophys 1993;307:153164.
32. Koob TJ, Clark PE, Hernez DJ, Thurmond FA, Vogel KG.
Compression loading in vitro regulates proteoglycan synthesis by tendon fibrocartilage. Arch Biochem Biophys 1992;
298:303312.
33. Cockburn CG, Barnes MJ. Characterization of thrombospondin binding to collagen (type I) fibers: role of collagen telopeptides. Matrix 1991;11:168176.
34. Amiel D, Gelberman R, Harwood F, Siegel D. Fibronectin in
healing flexor tendons subjected to immobilization or early
controlled passive motion. Matrix 1991;11:184189.
35. Kannus P, Jozsa L, Jarvinen TA, Jarvinen TL, Kvist M, Natri
A, Jarvinen M. Location and distribution of non-collagenous
matrix proteins in musculoskeletal tissues of rat. Histochem
J 1998;30:799810.
36. Jozsa L, Lehto M, Kannus P, Kvist M, Reffy A, Vieno T,
Jarvinen M, Demel S, Elek E. Fibronectin and laminin in
Achilles tendon. Acta Orthop Scand 1989;60:469471.
37. Mow VC, Mak AF, Lai WM, Rosenberg LC, Tang LH. Viscoelastic properties of proteoglycan subunits and aggregates in
varying solution concentrations. J Biomech 1984;17:325338.
38. Hooley CJ, Cohen RE. A model for the creep behaviour of
tendon. Int J Biol Macromol 1979;1:123132.
39. Kannus P. Structure of the tendon connective tissue. Scand J
Med Sci Sports 2000;10:312320.
40. Parry DA, Craig AS, Barnes GR. Tendon and ligament from
the horse: an ultrastructural study of collagen fibrils and
elastic fibers as a function of age. Proc R Soc London Ser B
Biol Sci 1978;203:293303.
41. Ross R. The elastic fiber. J Histochem Cytochem 1973;21:
199208.
42. Minns RJ, Soden PD, Jackson DS. The role of the fibrous
components and ground substance in the mechanical properties of biological tissues: a preliminary investigation. J Biomech 1973;6:153165.
43. Gosline JM. The elastic properties of rubber-like proteins and
highly extensible tissues. Symp Soc Exp Biol 1980;34:331357.
44. Evans JH, Nachemson AL. Biomechanical study of human
lumbar ligamentum flavum. J Anat 1969;105:188189.
45. Agarwal BD, Broutman LJ. Analysis and performance of
fiber composites New York: Wiley; 1980.
46. Gordon JE. The new science of stronmg materials. Harmondsworth: Penguin Books Ltd.; 1976.
47. Aspden RM. Fiber reinforcing by collagen in cartilage and
soft connective tissues. Proc R Soc London Ser B Biol Sci
1994;B-258:195200.
48. Goh KL, Aspden RM, Mathias KJ, Hukins DWL. Effect of
fiber shape on the stresses within fibers in fiber-reinforced
composite materials. Proc R Soc London Ser 1999;A-455:
33513361.
49. Goh KL, Mathias KJ, Aspden RM, Hukins DWL. Finite
element analysis of the effect of fiber shape on stresses in
an elastic fiber surrounded by a plastic matrix. J Mater Sci
2000;35:24932497.
50. Kelly A, Macmillan NH. Strong Solids. 3rd ed.; Oxford:
Oxford University Press; 1986.
51. DeVente JE, Lester GE, Trotter JA, Dahners LE. Isolation of
intact collagen fibrils from healing ligament [letter]. J Electron Microsc 1997;46:353356.
52. Holmes DF, Chapman JA, Prockop DJ, Kadler KE. Growing
tips of type-I collagen fibrils formed in vitro are near paraboloidal in shape, implying a reciprocal relationship between
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
accretion and diameter. Proc Natl Acad Sci USA 1992;89:

98559859.
Kadler KE, Holmes DF, Trotter JA, Chapman JA. Collagen
fibril formation. Biochem J 1996;316 (Pt 1): 111.
Aspden RM. The theory of fiber reinforced composite materials applied to changes in the mechanical properties of the
cervix during pregnancy. J Theor Biol 1988;130:213221.
Cabaud HE. Biomechanics of the anterior cruciate ligament.
Clin Orthop 1983; 2631.
Dodds JA, Arnoczky SP. Anatomy of the anterior cruciate
ligament: a blueprint for repair and reconstruction. Arthroscopy 1994;10:132139.
Noyes FR, DeLucas JL, Torvik PJ. Biomechanics of anterior
cruciate ligament failure: an analysis of strain-rate sensitivity and mechanisms of failure in primates. J Bone Joint Surg
Am 1974;56:236253.
Hukins DWL, Kirby MC, Sikoryn TA, Aspden RM, Cox AJ.
Comparison of structure, mechanical properties and functions of lumbar spinal ligaments. Spine 1990;15:787795.
Viidik A. Functional properties of collagenous tissues. Int
Rev Connect Tissue Res 1973;6:127215.
Claes L, Neugebauer R. In vivo and in vitro investigation of
the long-term behaviour and fatigue strength of carbon fiber
ligament replacement. Clin Orthop 1985;186:99111.
Tkaczuk H. Tensile properties of human lumbar longitudinal
ligaments. Acta Orthop Scand Suppl 1968; 115.
Shah JS, Jayson MIV, Hampson WGJ. Mechanical implications of crimping in collagen fibers of human spinal
ligaments. Proc Instn Mech Eng [H], J Eng Med 1979;8:
95102.
Kirby MC, Sikoryn TA, Hukins DWL, Aspden RM. Structures
and mechanical properties of the longitudinal ligaments and
ligamenta flava of the spine. J Biomed Eng 1989;11:192196.
Brand RA. Knee ligaments: a new view. J Biomech Eng
1986;108:106110.
Viidik A. Tensile strength properties of Achilles tendon systems in trained and untrained rabbits. Acta Orthop Scand
1969;40:261272.
Rigby J, Hirai N, Spikes JD, Eyring M. The mechanical
properties of rat tail tendon. J Gen Physiol 2003;43:265283.
Haut RC, Little RW. A constitutive equation for collagen
fibers. J Biomech 1972;5:423430.
Dorrington KL. The theory of viscoelasticity in biomaterials.
Symp Soc Exp Biol 1980;34:289314.
Cohen RE, Hooley CJ, McCrum NG. Viscoelastic creep of
collagenous tissue. J Biomech 1976;9:175184.
Kennedy JC, Hawkins RJ, Willis RB, Danylchuck KD. Tension studies of human knee ligaments. Yield point, ultimate
failure, and disruption of the cruciate and tibial collateral
ligaments. J Bone Joint Surg Am 1976;58:350355.
Neumann P, Keller TS, Ekstrom L, Perry L, Hansson TH,
Spengler DM. Mechanical properties of the human lumbar
anterior longitudinal ligament. J Biomech 1992;25:11851194.
Sikoryn TA, Hukins DWL. Mechanism of failure of the
ligamentum flavum of the spine during in vitro tensile tests.
J Orthop Res 1990;8:586591.
Azangwe G, Mathias KJ, Marshall D. Macro and microscopic
examination of the ruptured surfaces of anterior cruciate
ligaments of rabbits. J Bone Joint Surg Br 2000;82:450456.
Barry D, Ahmed AM. Design and performance of a modified
buckle transducer for the measurement of ligament tension. J
Biomech Eng 1986;108:149152.
Beynnon B, Howe JG, Pope MH, Johnson RJ, Fleming BC.
The measurement of anterior cruciate ligament strain in
vivo. Int Orthop 1992;16:112.
Woo SL, Newton PO, MacKenna DA, Lyon RM. A comparative evaluation of the mechanical properties of the rabbit
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
medial collateral and anterior cruciate ligaments. J Biomech

1992;25:377386.
Lam TC, Shrive NG, Frank CB. Variations in rupture site
and surface strains at failure in the maturing rabbit medial
collateral ligament. J Biomech Eng 1995;117:455461.
Sikoryn TA, Hukins DWL. Failure of the longitudinal ligaments of the spine. J Mater Sci Lett 1988;7:13451349.
Gupta P, Subramanian KN, Brinker WO, Gupta AN. Tensile
strength of canine cruciate ligaments inthe dog. Am J Vet Res
1971;32:183190.
Njus GO, Njus NM. A non-contact method for determining
cross-sectional area of soft tissue. Trans Orthopaed Res Soc
1984;32:126131.
Woo SL, Danto MI, Ohland KJ, Lee TQ, Newton PO. The use
of a laser micrometer system to determine the cross-sectional
shape and area of ligaments: a comparative study with two
existing methods. J Biomech Eng 1990;112:426431.
King GJ, Pillon CL, Johnson JA. Effect of in vitro testing over
extended periods on the low-load mechanical behaviour of
dense connective tissues. J Orthop Res 2000;18:678681.
Kwan MK, Lin TH, Woo SL. On the viscoelastic properties of
the anteromedial bundle of the anterior cruciate ligament. J
Biomech 1993;26:447452.
Provenzano P, Lakes R, Keenan T, Vanderby, Jr. R. Nonlinear ligament viscoelasticity. Ann Biomed Eng 2001;29:
908914.
Silver FH, Christiansen DL, Snowhill PB, Chen Y. Role of
storage on changes in the mechanical properties of tendon
and self-assembled collagen fibers. Connect Tissue Res 2000;
41:155164.
Woo SL, Gomez MA, Akeson WH. The time and historydependent viscoelastic properties of the canine medical collateral ligament. J Biomech Eng 1981;103:293298.
Thornton GM, Oliynyk A, Frank CB, Shrive NG. Ligament
creep cannot be predicted from stress relaxation at low stress:
a biomechanical study of the rabbit medial collateral ligament. J Orthop Res 1997;15:652656.
Thornton GM, Shrive NG, Frank CB. Altering ligament
water content affects ligament pre-stress and creep behaviour. J Orthop Res 2001;19:845851.
Speer KP, Warren RF, Wickiewicz TL, Horowitz L, Henderson
L. Observations on the injury mechanism of anterior cruciate
ligament tears in skiers. Am J Sports Med 1995;23:7781.
Fischer JF, Leyvraz PF, Bally A. A dynamic analysis of knee
ligament injuries in alpine skiing. Acta Orthop Belg
1994;60:194203.
Emerson RJ. Basketball knee injuries and the anterior cruciate ligament. Clin Sports Med 1993;12:317328.
Azangwe G, Mathias KJ, Marshall D. The effect of torsion on
the appearance of the rupture surface of the ACL of rabbits.
Knee 2002;9:3139.
Woo SL, Orlando CA, Gomez MA, Frank CB, Akeson WH.
Tensile properties of the medial collateral ligament as a
function of age. J Orthop Res 1986;4:133141.
Davison PF. The organization of collagen in growing tensile
tissues. Connect Tissue Res 1992;28:171.
Greenlee, Jr. TK, Pike D. Studies on tendon healing in the
rat. J Plast Reconstr Surg 1071;48:260270.
Matyas JR, Anton MG, Shrive NG, Frank CB. Stress governs
tissue phenotype at the femoral insertion of the rabbit MCL. J
Biomech 1995;28:147157.
Moriggl B, Kumai T, Milz S, Benjamin M. The structure and
histopathology of the enthesis organ at the navicular insertion of the tendon of tibialis posterior. J Rheumatol
2003;30:508517.
Cotton FJ, Morrison GM. Artifical ligaments at the knee: a
technique. N Engl J Med 1934;210:13311332.
251
99. Fujikawa K, Kobayashi T, Sasazaki Y, Matsumoto H, Seedhom BB. Anterior cruciate ligament reconstruction with the
Leeds-Keio artificial ligament. J Long Term Eff Med
Implants 2000;10:225238.
100. Chen CH, Chen WJ, Shih CH. Arthroscopic reconstruction of
the posterior cruciate ligament: a comparison of quadriceps
tendon autograft and quadruple hamstring tendon graft.
Arthroscopy 2002;18:603612.
101. Ruiz AL, Kelly M, Nutton RW. Arthroscopic ACL reconstruction: a 5-9 year follow-up. Knee 2002;9:197200.
102. Schroven IT, Geens S, Beckers L, Lagrange W, Fabry G.
Experience with the Leeds-Keio artificial ligament for anterior cruciate ligament reconstruction. Knee Surg Sports
Traumatol Arthrosc 1994;2:214218.
103. Drogset JO, Grontvedt T. Anterior cruciate ligament reconstruction with and without a ligament augmentation device:
results at 8-Year follow-up. Am J Sports Med 2002;30:851856.
104. Friedman MJ, Sherman OH, Fox JM, Del PW, Snyder SJ.
Ferkel RJ. Autogeneic anterior cruciate ligament (ACL)
anterior reconstruction of the knee. A review. Clin Orthop
Relat Res 1985;914.
105. Alexander H, Weiss AB. Editorial Comment. Clin Orthop
Relat Res 1985;23.
106. Fujikawa K, Ohtani T, Matsumoto H, Seedhom BB. Reconstruction of the extensor apparatus of the knee with the
Leeds-Keio ligament. J Bone Joint Surg Br 1994;76:200203.
107. Jenkins DH, Forster IW, McKibbin B, Ralis ZA. Induction of
tendon and ligament formation by carbon implants. J Bone
Joint Surg Br 1977;59:5357.
108. Turner IG, Thomas NP. Comparative analysis of four types of
synthetic anterior cruciate ligament replacement in the goat:
in vivo histological and mechanical findings. Biomaterials
1990;11:321329.
109. Marumo K, Kumagae Y, Tanaka T, Fujii K. Long-term
results of anterior cruciate ligament reconstruction using
semitendinosus and gracilis tendons with Kennedy ligament
augmentation device compared with patellar tendon autografts. J Long Term Eff Med Implants 2000;10:251265.
110. Fukubayashi T, Ikeda K. Follow-up study of Gore-Tex artificial ligamentspecial emphasis on tunnel osteolysis. J
Long Term Eff Med Implants 2000;10:267277.
111. Woo SL, Hildebrand K, Watanabe N, Fenwick JA, Papageorgiou CD, Wang JH. Tissue engineering of ligament and
tendon healing. Clin Orthop 1999; S312S323.
112. Huard J, Li Y, Peng H, Fu FH. Gene therapy and tissue
engineering for sports medicine. J Gene Med 2003;5:93108.
113. Gentleman E, Lay AN, Dickerson DA, Nauman EA, Livesay
GA, Dee KC. Mechanical characterization of collagen fibers
and scaffolds for tissue engineering. Biomaterials 2003;24:
38053813.
114. DesRosiers EA, Yahia L, Rivard CH. Proliferative and matrix
synthesis response of canine anterior cruciate ligament fibroblasts submitted to combined growth factors. J Orthop Res
1996;14:200208.
115. Martinek V, Latterman C, Usas A, Abramowitch S, Woo SL,
Fu FH, Huard J. Enhancement of tendon-bone integration of
anterior cruciate ligament grafts with bone morphogenetic
protein-2 gene transfer: a histological and biomechanical
study. J Bone Joint Surg Am 2002;84-A:11231131.
Further Reading
Akeson WH, Pedowitz R, OConnor JJ, editors. Knee Ligaments:
Structure, Function, Injury and Repair. 2nd ed. New York:
Lippincott Williams & Wilkins; 2003.
Mow VC, Hayes WC, editors. Basic Orthopaedic Biomechanics.
New York: Raven Press; 1991.
252
Evans CH, Scully SP, guest editors. Orthopaedic Gene Therapy.

Clinical Orthopaedics and Related Research. Volume 379
Suppl., 2000.
See also BONE
PROPERTIES OF.
AND TEETH, PROPERTIES OF; CARTILAGE AND MENISCUS,
LINEAR VARIABLE DIFFERENTIAL

TRANSFORMERS
SUNIL KESAVAN
Akebono Corporation
Farmington Hills, Michigan
NARENDER REDDY
The University of Akron
Akron, Ohio
INTRODUCTION
Several methods of transduction are available to convert
physiological events into electrical signals. Basic physiological variables are first converted by sensing elements
into variables that can easily be measured by available
transducers. One such transducer, the linear variable
differential transformer, commonly abbreviated as LVDT
(some manufacturers designate it as LDVT linear differential voltage transformer), is used to convert mechanical
displacement into proportional electronic signals. LVDTs
are capable of measuring physiological variables, such as
displacement, pressure, force, and acceleration, which are
either available in the form of a linear displacement or can
be converted into such movement.
Figure 1. Schematic of the cutaway view of an LVDT showing the

core, primary coil, and two secondary coils: (ei) excitation voltage;
(e0) output voltage.
nection in the LVDT causes the phase of the output voltage

to change by 1808 as the core passes through the null
position. The output voltage, e0, is generally out of phase
with the excitation voltage, ei. The phase shift is dependent
on the frequency of ei, and each LVDT has a particular
frequency at which phase shift is zero.
FABRICATION
The LVDT features essentially frictionless measurement
and long mechanical life, because there is no mechanical
THEORY
An LVDT is an inductive electromechanical transducer
that uses a primary (energizing) coil and two seriesopposed secondary coils. This mode of connecting the
secondaries serves to mutually cancel out the secondary
voltages. In this popular configuration, due to Shaevitz (1),
the primary winding is symmetrically placed with respect
to the secondary windings on a cylindrical former. The
former surrounds a free-moving rod-shaped magnetic core,
which provides a path for the magnetic flux linking the
coils (Fig. 1). The magnetic core is connected to a sensing
device like a movable diaphragm. Movement of the sensor
induces core movement, which in turn produces voltage
variations that are measured directly.
When the sliding magnetic core is in the central (null)
position, the electromotive forces (emfs) generated in the
secondaries are equal, and the net output voltage, e0 is,
therefore, zero. Movement of the core from this central
position causes the mutual inductance (coupling) for one
coil to increase and the other coil to decrease. The amplitude of the output voltage, e0, being the difference between
the emfs in the two secondaries, varies approximately
linearly with the position of the core on either side of
the null position (Fig. 2). The differential secondary con-
Figure 2. Output voltage of an LVDT as a linear function of core

position.
contact between the enclosed coil assembly and the separate freely moving core within the coil assembly (Fig. 1).
A typical alternating current (ac) LVDT core consists of
a uniformly dense cylindrical slug of a high permeability
nickeliron alloy. The core is internally threaded to accept
nonmagnetic core rods from an external sensing or actuating element. The core moves within a cylindrical coil
assembly. Hollow cores are employed when a low mass
core is desired. Recently, researchers have developed lightweight glass-covered amorphous wire cores that can be
used to fabricate high sensitivity LVDTs with good
mechanical and corrosion resistance (2). The primary
and secondary windings are spaced symmetrically by winding them on a slotted cylindrical former. To avoid material
corrosion or insulation leakage, the windings are impregnated with an insulating varnish under a vacuum. The
coils are encapsulated in epoxy for further mechanical and
moisture protection. Magnetic or ferromagnetic materials
in the proximity of an LVDT can disrupt its magnetic field.
The magnetic field of an LVDT may also induce eddy
currents into nonmagnetic materials in its vicinity. These
currents in turn would create a magnetic flux that would
interfere with the LVDT output. These problems are
avoided in practice by enclosing the LVDT in a case fabricated from an alloy, a high permeability iron, or a stainless steel. The LVDT assembly is then mounted in a C- or
split block. LVDTs that can measure rotational movement
are also available.
In addition to the ac LVDTs described in the previous
sections, direct current (dc) LVDTs are also available (3,4).
These LVDTs, in addition to having all the advantages of ac
LVDTs, possess the simplicity of dc operation. They consist
of two integral parts: an ac-operated LVDT and a carrier
generatorsignal conditioning module. The small carrier
system eliminates the need for the ac excitation, demodulation, and amplification equipment required for conventional ac LVDTs. This cuts down the cost and reduces the
volume of LVDT instrumentation; dc units can be battery
operated or be supplied by a simple dc power supply (3,4).
Also, any dc meter can be employed to read the LVDT
output. These advantages, coupled with the small size of
the dc LVDTs, make them attractive for use in hospitals
and other medical environments.
The LVDTs have several advantages and a few disadvantages (36) as briefly reviewed next.
1. Essentially frictionless operation and long mechanical life: As described in the previous section, the
LVDT has no moving mechanical contact between
the moving core and the windings. This ensures that
LVDTs have a fast dynamic response as no additional
load apart from the core mass is imposed on the
measured event. In addition, this helps LVDTs to
have a long, essentially infinite, mechanical life.
2. Good in hostile environments: LVDTs can be manufactured to withstand the vagaries of chemical
corrosion and extremes of temperature and pressure. This is facilitated by the separation between
the core and windings of the LVDT. Only a static
seal is required to isolate the coil assembly from
hostile environments.
253
3. Extremely high resolution: LVDTs can respond to

extremely small displacements. Microdisplacement
LVDT transducers capable of measuring displacements down to 100 pm have been fabricated (7).
4. Null repeatability: The null position of an LVDT is
very repeatable, even with large temperature variations.
5. InputOutput isolation: Since the primary and secondary windings are isolated from each other, the
signal ground can be isolated from the excitation
ground.
6. Cross-axis rejection: The LVDT is only responsive to
axial core motion. Cross-axis motion induced by conditions such as jarring or continuous vibration will
not affect the LVDT output.
7. Overtravel damage resistance: As the LVDT core
can pass completely through the coil assembly, the
transducer is inherently immune to damage from
unanticipated overtravel that can be encountered in
applications where materials or structures can yield
or fail.
8. Absolute output: Unlike a lot of other transducers
that are incremental output devices, LVDTs are
absolute output devices, that is, the displacement
information from an LVDT is not lost if the system
loses power. When the measuring system is restarted,
the LVDTs output value will be the same as it was
before the power failure occurred.
All these advantages, in addition to their reasonable
cost, have made the LVDT an attractive displacement
measurement technique. However, LVDTs for use in medical applications have the following disadvantages: (1)
They require a constant amplitude excitation of high frequency. (2) They cannot be used in the vicinity of equipment that creates strong magnetic fields.
LVDT INSTRUMENTATION
Instrumentation normally used with an ac LVDT should
perform the following functions (3,4).
Excitation
An LVDT needs an ac input of constant amplitude at a
frequency that is not readily available. Hence, an oscillator
of the appropriate frequency has to be connected to an
amplifier with amplitude regulation on its output.
Amplification
As in the case of most transducers, the low level outputs of
LVDTs require amplification. One procedure for amplification employs two steps: (1) use of an ac carrieramplifier
before demodulation; and (2) a dc amplifier after the demodulator (3,4).
Demodulation
As discussed earlier, the output of an LVDT remains
proportional to the displacement while it undergoes a
254
Figure 3. Block diagram for an LVDT employing a series 1000 oscillatordemodulator supplied by
Trans-Tek, Inc. (Courtesy of Trans-Tek Inc.)
phase shift of 1808 when the core goes through the null.
When this LVDT output is connected to a voltmeter, the
meter will register the same reading for equal amounts of
core displacement on either side of the null position. This
lack of directional sensitivity has to be overcome if one has
to tell to which side of the null the core is displaced. Two
techniques can be used to confer directional sensitivity on
an LVDT output. In one technique, the core is offset, and
the operation is centered on a position other than the null
point. In this case, the output signal either increases or
decreases. The other procedure uses a phase-sensitive
demodulator (also called a detector). Several such devices
are available and are discussed in detail elsewhere (8). The
simplest forms employ diode rectification while the complex forms involve synchronous demodulation. Figure 3
shows the block diagram for an LVDT employing a series
1000 oscillatordemodulator supplied by Trans-Tek, Inc. (9).
The demodulator confers directional sensitivity on its
input (output of the LVDT), which is either in phase with,
or 1808 out of phase with, the carrier signal (10). The
demodulator output e0 is usually sent to a low pass filter
that will pass only the frequencies present in x and reject
all higher frequencies created by the modulation procedure. Obviously, demodulation is not required if the LVDT
transducer is to be used only on one side of the null position.
Recent developments allow all LVDT support circuitry
to be accomplished using an inexpensive flexible field programmable analog array (FPAA). The FPAA consists of
configurable analog blocks consisting of switched-capacitor op-amp cells surrounded by a programmable interconnect and I/O structure (11).
dc Power
Stable dc voltage sources are required for operation of the
electronics associated with LVDTs. The dc LVDTs available at the present time employ a microcircuit module
including all the electronics needed to provide ac excitation
to the primary of the LVDT and to demodulate and amplify
the analog LVDT signal. The module is mounted in tandem
with the LVDT and only increases the effective LVDT
length slightly.
Figure 4 shows the block diagram for a dc LVDT (9). The
oscillator produces a constant amplitude sine wave
excitation for the primary of the LVDT. A phase sensitive
demodulator and an RC filter network process the secondary coil output. Some dc LVDT modules are furnished with
a reverse polarity protector for the dc power input. dc LVDTs

are becoming increasingly popular due to their advantages in the areas of calibration and signal conditioning.
SELECTION CRITERIA
Several criteria have to be considered in selecting a particular LVDT for a certain application (12). The manufacturer supplies several of these parameters as specification
criteria.
Total Stroke
Stroke-length specification in the selection of an LVDT for
a particular application is governed by the displacement to
be measured. LVDTs can be custom-made for either short(up to 0.01 m) or long-stroke (up to 1.5 m) operation;
however, cost of fabrication increases greatly with increase
in length, and lengths over 0.03 m may not be cost effective.
Linearity and the Nominal Linear Range
The output of an LVDT is a nearly linear function of core
position for a rather wide range on either side of the
balance (null) position (Fig. 2). A nominal linear range is
defined for an LVDT as the core displacement on either side
of the balance position for which the LVDT output as a
function of displacement remains a straight line. Outside
this range, the output starts to deviate gradually from the
ideal straight line in the form of a smooth curve. Linearity
of an LVDT is defined as the maximum deviation from a
best-fit straight line (applied to a plot of LVDT output
voltage vs. core displacement) within the nominal linear
Figure 4. Block diagram of a dc LVDT. (Courtesy of Trans-Tek

Inc.)
range (12). Linearity is usually expressed as a percentage

of the full scale. A typical LVDT has a linearity of about
0.25%.
255
original position is an important consideration. LVDTs

with repeatability better than 100 nm are available for
some critical applications.
Sensitivity
The sensitivity of an LVDT is usually expressed as the
output in millivolts (or V) per 0.001 m core displacement
per volt input. Normally, both the input voltage and the
frequency are specified as well, because voltage sensitivity
may vary with frequency over a limited frequency range. A
typical miniature LVDT transducer has a sensitivity of
8200 mV out/0.001 m/V input.
Resolution
Resolution of an LVDT is the smallest core movement that
can produce an observable change in the voltage output
(12). With careful circuit design, displacements smaller
than 100 nm can be detected.
Armature Mass
The mass of the armature (core) of the LVDT should be
small so as not to unduly load the measured event. A
reduction in the length of the LVDT results in a reduction
in either the linearity or the maximum linear range,
whereas sensitivity increases.
Excitation Frequency and Voltage
The sensitivity of the LVDT depends on both the excitation
voltage and frequency. Normally, a sinusoidal voltage of
315 V rms amplitude and a frequency of 60 Hz20 kHz is
used for the excitation of LVDTs. The sensitivity of an
LVDT increases with the excitation frequency, particularly
at the lower part of the operating frequency range (12).
Normally, an excitation frequency range of 15 kHz produces optimal LVDT operation.
Operating Environment
LVDTs have the advantage of being available in hostileenvironment-proof format. Transducers designed to withstand both high and cryogenic temperatures and high
pressures are available. Immersion-type LVDTs resistant
to corrosive liquids are also available. Normally, specification criteria for an LVDT include information on the temperature range of operation and the temperature coefficient.
Residual Voltage Output
The residual voltage output is the LVDT output when the
core is in the null position. This should ideally be zero;
however, the null voltages and the harmonics of the excitation source do not cancel, resulting in a nonzero residual
output (12). In practice, the residual voltage is about 1% of
that obtained with maximum displacement.
Repeatability
Repeatability, the ability of the LVDTs to give the same
output if the core is displaced and returned to the
MEDICAL APPLICATIONS
LVDTs are used in medical applications and research to
measure physiological variables that are either available in
the form of a linear displacement or can be converted into
such movement. LVDTs for medical applications can be
readily fabricated in very small sizes with low mass cores.
This will ensure that only a negligible force is imposed on
the measured physiological event. Also, due to the low
alternating currents in the windings, negligible magnetic
load is imposed. When not in use, the core remains in the
null position, and no force is imposed on the measured
event. Even when the core is displaced from null, the load
imposed on the event is small. These advantages, coupled
with the general advantages of LVDTs discussed in the
previous paragraphs, make these transducers very attractive for physiological measurements.
One early application of LVDTs was in the fabrication of
invasive blood pressure measurement transducers (13).
These transducers consisted of three essential parts: (1)
a dome with pressure fittings, (2) a stainless steel diaphragm and core assembly, and (3) the LVDT coils. Pressure transmitted via the catheter exerts a force on the
diaphragm. This causes a movement of the diaphragm,
which in turn manifests itself in a movement of the core
attached to it. Movement of the core of the LVDT creates a
proportional output that can then be recorded after suitable electronic circuit processing. Catheter tip and implantable transducers employed the same principle (13).
However, these rugged LVDT blood pressure transducers
have been supplanted by cost-effective microelectromechanical system (MEMS) type transducers (14).
Another application for LVDT transducers is in indentation tests on tissue to determine mechanical properties.
The authors have developed an LVDT indenter for the
characterization of the mechanical properties of skin and
the underlying soft tissue (Fig. 5). The indenter uses a
loaded hemispherical tip coupled with a load cell-LVDT
system for simultaneously measuring both the force and
displacement during indentation tests. This information in
turn was used to evaluate soft tissue properties. Walsh and
Schettini (15) used a similar indenter to measure the
in vivo viscoelastic response properties of brain tissue.
Oculotonometers operating on the same principle and
designed to indent the corneoscleral shell use LVDTs to
measure deflections in the micrometer range (16). In
another similar application, Gunner et al. (17) used an
LVDT transducer-mounted extensometer to measure the
in vivo recoil characteristics of human skin. The device
consisted of two flat rectangular tabs, one fixed and
the other capable of rectilinear sliding motion, attached
to the test skin surface with double-sided adhesive tape.
This combination was attached to an LVDT displacement
transducer. Behavior of the skin resulting from the movement of the tabs was converted by the LVDT into electronic
signals that were then analyzed to characterize the skin.
256
Figure 6. Mechanism of modular brain probe drive showing the

LVDT used to help in precise electrode placement. [Courtesy of
Radionics (a division of Tyco Healthcare).]
Figure 5. An LVDT skin and tissue indenter. (Courtesy of N. P.

Reddy, J. Kagan and G. V. B. Cochran.)
Christiansen et al. (18) used a similar device for the

viscoelastic characterization of skin. These examples are
just a few of the myriad potential applications for LVDT
transducers in soft tissue characterization.
Several radiological and neurological devices have
incorporated LVDTs. For example, Laser Diagnostic
Technologies of San Diego, CA, incorporated a positionsensing DC-DC LVDT into a scanning laser tomography
instrument designed for retinal topography (9). The stable
and repeatable DCDT output is part of a continuous
feedback loop in the scanners on-board logic control system. Radionics, Inc., employed an LVDT in a sophisticated
modular probe drive used to support the precise implantation of deep brain stimulating electrodes (9). The device
uses a pushpull cable drive mechanism to move the
carrier that guides the probe to the desired location in
the brain. The LVDT is mounted at the top of the mechanism and is used to accurately monitor probe position
(Fig. 6). The LVDT used in this application was sealed to
resist moisture and was modified to withstand the rigors
of steam sterilization.
LVDTs have been used in endocrinology and pharmacology to evaluate in vitro and in vivo contractile properties
of vascular smooth muscle. Erdos et al. (19) designed an
in vitro isotonic myograph employing an LVDT. The

device resembled a beam-type balance. One arm of the
device was connected to the contracting muscle specimen,
and the other arm was counterbalanced by a suspended
weight. Motion of the weight was translated via the movement of an LVDT core into electronic signals.
Gow (20) employed a novel LVDT electronic caliper for
the continuous monitoring of arterial diameter changes
during pulsation. This low mass device was found to
possess a rapid response time with a natural resonant
frequency greater than 180 Hz. Shykoff et al. (21) used
LVDT measurements of changes in diameter of dorsal
hand veins to establish diameter, pressure, and compliance relationships. LVDT-based devices have been used to
evaluate the in vivo vascular effects of drugs with the
dorsal hand vein technique (22). For example, Landau et
al. (23) used an LVDT-based device to evaluate the magnesium-induced vasodilation in the dorsal hand vein. A
similar technique was used by Streeten and Anderson (24)
to measure venous contractile responses to locally infused
norepinephrine.
LVDTs have been used in numerous orthopedic and
dental devices. Chen et al. (25) used an LVDT bonded to
the mandibular first molars to quantify mandibular deformation during mouth opening. Other possible medical
applications are the mapping of facial contours before
and after maxillofacial surgery and the profiling of spinal
deformation in abnormalities like scoliosis. Buhler et al.
(26) and Flamme et al. (27) used LVDTs to quantify
micromotion in orthopedic implants. Recently, Dong et
al.(28) incorporated an LVDT into a device for quantitative
assessment of tension in wires of fine-wire external orthopedic
fixators.
An LVDT was used to calibrate finger movements and to
correlate these movements with surface electromyograms
from the flexor digitorum superficialis muscles in the
forearm (29,30). This work was designed to develop techniques for control of anthropomorphic teleoperator fingers
using surface electromyographic signals obtained from the

forearm.
Wang et al. (31) used six spring-loaded LVDTs in an
experimental technique to measure three-dimensional (3D),
six-degrees-of-freedom motion of human joints. Rotary
LVDTs are useful for measuring joint angles. For measuring
3D rotations, rotary LVDTs are incorporated into sixdegree-of-freedom motion linkages.
As illustrated in the applications discussed above, LVDTs
are highly suited for biomedical device and research applications requiring accurate displacement measurements
with high resolution, inputoutput isolation, and crossaxis rejection. Although LVDTs are being replaced by
miniaturized, cost-effective transducers utilizing advanced
fabrication technologies in many applications, their advantages still render them excellent candidates for biomedical
applications.
ACKNOWLEDGMENT
The authors would like to acknowledge the help provided
by Rema Menon in the preparation of this manuscript.
BIBLIOGRAPHY
1. Schaevitz H. The linear variable differential transformer.
Proc Soc Stress Anal 1947;4:7988.
2. Chiriac H, Hristoforou E, Neagu M, Pieptanariu M. Linear
variable differential transformer sensor using glass-covered
amorphous wires as active core. J Magn Magn Mater 2000;
215:759761.
3. Schaevitz Engineering. Technical bulletins 1002D and 7007.
Pennsauken (NJ) 1986.
4. Schaevitz Sensors. Shaevitz Sensor Solutions. Catalog No.
SCH-2001 Hampton (VA) 2000.
5. [Anonymous]. No date. An LVDT Primer. [Online]. Macro
Sensors. Available at www.macrosensors.com. 2005 March 8.
6. Weinstein E. LVDTs on the factory floor. Instrum Control
Syst 1982;55:5961.
7. Sydenham PH. Microdisplacement transducers. J Phys E
1972;6:721733.
8. Szczyrbak J, Schmidt EDD. LVDT signal conditioning techniques. Meas Control 1997;183:103111.
9. [Anonymous]. No date. LVDT application. [Online]. TransTek Inc. Available at www.transtekinc.com. 2005 March 8.
10. Doebelin EO. Measurement Systems: Application and
Design. 4th ed. New York: McGraw-Hill; 1990.
11. Severn J, October. 2001. New analog interface for LVDTs.
[Online]. Industrial Technology. Available at www.industrialtechnology.co.uk. 2005 March 8.
12. Anonymous, Finding the right LVDT. Instrum Control Syst
1977;50:6162.
13. [Anonymous]. No date. LVDT Applications. [Online]. Macro
Sensors. Available at www.macrosensors.com. 2005 March 8.
14. Seeley RS. 1996. The future of medical microelectromechanical systems. [Online]. Available at www.devicelink.com/
mem/archive/96/01/003.html. 2005 March 8.
15. Walsh EK, Schettini A. A pressure-displacement transducer
for measuring brain tissue properties in vivo. J Appl Physiol
1975;38:187189.
16. Stepanik J. The Mackay-Marg Tonometer. Acta Ophthal
1970;48:1140.
257
17. Gunner CW, Hutton WC, Burlin TE. An apparatus for measuring the recoil characteristics of human skin in vivo. Med
Biol Eng Comput 1979;17:142144.
18. Christiansen MS, Hargens III CW, Nacht S, Gans EH. Viscoelastic properties of intact human skin: Instrumentation,
hydration effects, and the contribution of the stratum corneum. J Invest Dermatol 1977;69:282286.
19. Erdos EG, Jackman V, Barnes WC. Instrument for recording
isotonic contractions of smooth muscles. J Appl Physiol
1962;17: 307308.
20. Gow BS. An electrical caliper for measurement of pulsatile
arterial diameter changes in vivo. J Appl Physiol 1966;21:
11221126.
21. Shykoff BE, Hawari FI, Izzo JL. Diameter, pressure and
compliance relationships in dorsal hand veins. Vasc Med
2001;6(2):97102.
22. Pang YC. Autonomic control of venous system in health and
disease: effect of drugs. Pharmacol Therapeut 2001;90:179
230.
23. Landau R, Scott JA, Smiley RM. Magnesium-induced vasodilation in the dorsal vein. BJOG (Bri J Obst Gyn) 2004;111:
446451.
24. Streeten DHP, Anderson GH. Mechanisms of orthostatic
hypotension and tachycardia in patients with pheochromocytoma. AJH 1996;9:760769.
25. Chen DC, Lai YL, Chi LY, Lee SY. Contributing factors of
mandibular deformation during mouth opening. J Dent
2000;28(8):583588.
26. Buhler DW, Oxland TR, Nolte LP. Design and evaluation of a
device for measuring three-dimensional motions of press-fit
femoral stem prosthesis. Med Eng Phys 1999;19:187199.
27. Flamme CH, Kohn D, Kirsch L, Hurschler C. Primary stability of different implants used in conjunction with high
tibial osteomy. Arch Orth Trauma Surg 1999;119:450455.
28. Dong Y, Saleh M, Yang L. Quantitative assessment of tension
in wires of fine-wire external fixators. Med Eng Phys
2005;27:6366.
29. Gupta V, Reddy NP. Surface electromyogram for the control
of anthropometric teleoperator fingers. Weghorst SJ, Soeburg HB, Morgan KS, editors. Medicine Meets Virtual Reality: Healthcare in the Information Age. Amsterdam: IOP
Press; 1996.
30. Devavaram A, Reddy NP. Intelligent systems for control of
telemanipulators using surface EMG signals, submitted for
publication.
31. Wang M, Bryant JT, Dumas GA. A new in vitro measurement technique for small three-dimensional joint motion and
its application to the sacroiliac joint. Med Eng Phys
1996;18(6): 495501.
Further Reading
Herceg ED. Handbook of Measurement and Control, Pennsauken
(NJ): Schaevitz Engineering; 1972.
Anonymous. LVDTs remain State-of-the-art. Meas Inspect Technol 1982;4(2):1316.
Anonymous. Displacement transducers (linear variable differential transformer products review). Control Instrum 1984;16(8):
2325.
Geddes LA, Baker LE. Principles of Applied Biomedical Instrumentation. 3rd ed. New York: Wiley-Interscience; 1989.
Webster JG. Medical Instrumentation: Application and Design.
3rd ed. New York: John Wiley & Sons; 1998.
See also BLOOD
PRESSURE
TEMPERATURE SENSOR.
MEASUREMENT ;
INTEGRATED
CIRCUIT
258
LITHOTRIPSY
LITERATURE, MEDICAL PHYSICS. See MEDICAL

PHYSICS LITERATURE.
LITHOTRIPSY
ALON Z. WEIZER
GLENN M. PREMINGER
Duke University Medical Center
Durham, North Carolina
INTRODUCTION
The clinical introduction of shock wave lithotripsy (SWL)
by Chaussy in 1980 has revolutionized the way in which
patients with renal and ureteral calculi are treated. Shock
wave lithotripsy is a noninvasive method of fragmenting
stones located inside the urinary tract. Since its initial
introduction, SWL technology has advanced rapidly in
terms of the means for shock wave generation, shock wave
focusing, patient coupling, and stone localization. Despite
rapid technological advances, most current commercial
lithotripters are fundamentally the same; they produce a
similar pressure waveform at the focus, which can be
characterized by a leading shock front with a compressive
wave followed by a trailing tensile wave (Fig. 1). The
acoustic fields produced by different lithotripters differ
from each other in terms of the peak amplitudes of the
pressure waveform, pulse duration, beam size, total acoustic energy, and therefore, their overall performance.
Clinical experience has guided the technical development
of second and third generation lithotripters with the aim of
providing user convenience and multifunctionality of the
device, rather than on further understanding of how SWL
fragments calculi or injures surrounding renal tissue.
Furthermore, the evolution of lithotripter design thus far
has overwhelmingly relied upon the importance of the compressive wave component of the shock wave (positive portion
of the sound wave), with almost total neglect of the contribution of the tensile component of the waveform. Consequently,
current lithotripters have suffered from inferior fragmentation rates compared to the original HM3 lithotripter.
In contrast, significant progress in SWL basic science
research has been made in the past 5 years to improve our
understanding of the primary mechanisms for stone comminution (fragmentation) and tissue injury. It is now
recognized that the disintegration of renal calculi in a
lithotripter field is the consequence of dynamic and synergistic interaction of two fundamental mechanisms: stress
wave-induced dynamic fracture in the form of nucleation,
growth, and coalescence of preexisting microcracks inside
the stone (1) and cavitation erosion caused by the violent
collapse of bubbles near the stone surface (2,3). Similarly,
two different mechanisms have been proposed for SWLinduced tissue injury: shear stress due to shock front
distortion (4) and cavitation induced inside blood vessels,
especially the expansion of intraluminal bubbles (5).
To understand how SWL fragments stones and causes
tissue injury, the basic components of current lithotripters,
the mechanisms behind stone fragmentation and kidney
injury, and clinical results of the original electrohydraulic
lithotripter in fragmenting kidney stones in patients will
be described. In addition, the future directions of SWL will
be reviewed based on current research that is investigating
ways to make lithotripters more efficient and safer.
HISTORY AND EVOLUTION OF SWL
Physicists at Dornier Systems, Ltd. and Friedrich Shafen,
Germany began experimenting with shock waves and their
travel through water and tissue in 1963. Throughout the
1970s, numerous experimental lithotripters were developed that used new methods of shock wave generation
and focusing as well as different techniques of stone localization. In addition, experimental studies were being performed in vitro and in vivo (in animal models) examining
the effects of shock waves on various organs and tissues.
In 1980, Chaussy and associates successfully treated the
first human and reported their first series of 72 patients in
1982 (6). Subsequently, > 1800 articles have been published
in the peer-reviewed literature, detailing the use of SWL for
the management of renal and ureteral calculi. Moreover,
numerous second and third generation devices have been
introduced and are currently being used throughout the
world. To understand how SWL results in stone fragmentation, the fundamentals of this technology are reviewed.
SWL PRINCIPLES
Figure 1. Pressuretime relationship of typical shock waves.

Typical pattern of a lithotripter-generated shock wave. The shock
wave is characterized by a leading positive or compressive
component (P) followed by a negative or tensile component (P).
Despite the tremendous number of lithotripters currently

available for fragmentation of renal and ureteral stones, all
of these devices rely on the same laws of acoustic physics.
Shock waves (i.e., high pressure sound waves) consisting of
a sharp peak in positive pressure followed by a trailing
negative wave are generated extracorporeally and passed
through the body to fragment stones. Sounds waves readily
propagate from a water bath or water medium into the
human body, due to similar acoustic impedances.
As a consequence, all lithotripters share four main
features: an energy source to generate the shock wave, a
LITHOTRIPSY
2a
Kidney stone
(F2)
2b
259
Kidney stone
(F2)
Ellipsoid reflector
and fluoroscopy
(focusing)
Water Bath
(coupling)
Spherical focusing of
piezo-ceramic elements
(F1)
Spark electrode (F1) (shock wave)
2c
2d
Kidney stone
(F2)
Kidney stone
(F2)
Parabolic
reflector
(focusing)
Acoustic
lens
(focusing)
Electromagnetic coil (F1) vibrates

sending out shock wave
Electromagnetic cylinder
vibrates (F1)
Figure 2. Schematic of different shock wave lithotripters. (2a) Electrohydraulic lithotripsy. Spark
electrode generates shock wave which is focused at F2 by an ellipsoid reflector. In the original
electrohydraulic lithotripter, a water bath served as a coupling medium to allow passage of the
shock wave from the source (F1) to the patient and stone (F2). Fluoroscopy or ultrasound can be used
to focus to identify and focus the shock waves on the stone. 2b) Piezoelectric lithotripsy. Many ceramic
elements are arranged in a spherical pattern. When energy passes through them, they vibrate and
send shock waves through a coupling medium to the stone. 2c and d) Electromagnetic lithotripsy. An
electromagnetic coil (2c) or cylinder (2d) are stimulated to vibrate by passage of electric current. Their
shock waves are focused on the stone (F2) by an acoustic lens (2c) or a parabolic reflector.
device to focus the shock wave at a focal point, a coupling

medium, and a stone localization system. (Fig. 2a) The
original electrohydraulic lithotripter utilized a spark plug
energy generator with an elliptical reflector for focusing
the shock waves. A water bath transmitted the shock
waves to the patient with stone localization provided by
biplanar fluoroscopy. Modification of the four basic components of this first generation lithotripter has provided the
development of second and third generation devices that
are currently available.
Shock Wave Generation and Focusing
While all lithotripters share the four aforementioned features,
it is the mode of shock wave generation that determines the
actual physical characteristics of that particular device. The
types of energy sources differ on how efficient they are in
fragmenting stones and how many treatments are required to
adequately treat the stone. Figure 2 diagrammatically
summarizes the three different shock wave generation

sources used in commercially available lithotripters.
Electrohydraulic Generator. In the original Dornier
HM3 lithotripter, the electrohydraulic generator (Fig. 2a)
was located at the base of a water bath and produced shock
waves by electric spark discharges of 15,00025,000 V of
1 ms duration. This high voltage spark discharge produced
the rapid evaporation of water that created a shock wave by
expanding the surrounding fluid. The generator was located
in an ellipsoidal reflector that concentrated the reflected
shock waves at a second focal point, F2, with F1 being the
point of origin of the primary shock waves. While the HM3
continues to be the gold standard lithotripter for stone
fragmentation, the short half-life of its electrode results
in variable pressure between shocks the longer it is used.
In addition, minimal displacement of the electrode, as it
deteriorates at F1, results in displacement of the F2
resulting in inaccurate focusing of the shock wave on the
260
LITHOTRIPSY
stone. The need for frequent replacement of the electrode

increases the cost of electrohydraulic lithotripsy.
Piezoelectric Generator. Piezoelectric shock waves are
generated (Fig. 2b) by the sudden expansion of ceramic
elements excited by a high frequency, high voltage pulse.
Thousands of these elements are placed along the inner
surface of a hemisphere at the base of a pool of water. While
each of these elements moves only slightly in response to a
pulse of electrical energy, the summation of the simultaneous expansion of multiple elements results in a high
energy shock wave directed to the focal area, at the center
of the sphere. The shock wave is propagated through either
a small water basin or a water-filled bag to the focal point,
F2. The spherical focusing mechanism of the piezoelectric
lithotripters provides a wide area of shock wave entry at
the skin surface, which causes minimal patient discomfort,
but a very small focal area with the smallest amount of
energy at F2 compared to other energy sources (7). The
small focal area necessitates greater precision to focus and
fragment the stone inside the kidney.
Electromagnetic Generator. In electromagnetic devices
(Fig. 2c and d), shock waves are generated when an electrical impulse moves a thin, spherical metallic membrane,
which is housed within a cylindrical shock tube. The resulting shock wave, produced in the water-filled shock tube,
passes through an acoustic lens and is thereby directed to
the focal point, F1 (Fig. 2c). The shock wave is coupled to the
body surface with a moveable water cushion and coupling
gel (8). Alternatively, when energy is passed through a
cylindrical coil, the resulting magnetic field pushes away
the surrounding cylindrical membrane producing a shock
wave that can be focused by a parabolic reflector (Fig. 2d).
While these devices produce reliable shock waves with
consistent pressures and focus on F2, they also produce
small focal regions that may result in reduced stone fragmentation and higher tissue parenchymal injury.
Shock Wave Focusing. Shock waves must be focused in
order to concentrate their energy on a calculus. The type of
shock wave generation dictates the method of focusing
used. Machines that utilize point sources, such as sparkgap electrodes (electrohydraulic lithotripters), generate
shock waves that travels in an expanding circular pattern.
All of these machines use ellipsoid reflectors for focusing
shock waves at the second focal point, F2.
Since a single piezoelement produces a very small
amount of energy, larger transducers with multiple ceramic elements are required for piezoelectric lithotripters.
The array of elements is positioned in a spherical dish that
allows focusing in a very small focal region, F1. Finally, the
vibrating metal membranes of the electromechanical lithotripters produce an acoustical plane wave that uses an
acoustic lens for focusing the shock wave at F1.
Coupling Medium
The original Dornier HM3 machine utilized a 1000 L water
bath to transmit shock waves into the patient. This method
of patient coupling required unique positioning of the
patient, since the anesthetized subject had to be lowered
into the tub and the calculus accurately positioned at the

second focal point. Second generation lithotripters were
designed to alleviate the physiologic, functional, and economic problems of a large water bath. Current models utilize
an enclosed water cushion, or a totally contained shock tube,
to allow simplified positioning and dry lithotripsy (9).
Stone Localization
Stone localization during lithotripsy is accomplished with
either fluoroscopy or ultrasonography. Fluoroscopy provides the urologist with a familiar modality and has the
added benefits of effective ureteral stone localization. However, fluoroscopy requires more space, carries the inherent
risk of ionizing radiation to both the patient and medical
staff and is not useful in localizing radiolucent calculi
(certain noncalcium-containing stones are not seen on
fluoroscopy or conventional radiographs).
Ultrasonography utilizes sound waves to locate stones.
These sound waves are generated at a source. When they
encounter different tissue densities, part of the sound wave
is reflected back and these reflected waves are used to
generate a two dimensional image that can be used to
focus the shock wave on a calculus. Sonography-based
lithotripters offer the advantages of stone localization with
continuous monitoring and effective identification of radiolucent stones, without radiation exposure (7). Additionally, ultrasound has been documented to be effective in
localizing stone fragments as small as 23 mm, and is as
good or better than routine radiographs to assess patients
for residual stone fragments following lithotripsy (8). The
major disadvantages of ultrasound stone localization
include the basic mastery of ultrasonic techniques by the
urologist and the inherent difficulty in localizing ureteral
stones. While there are several systems that utilize both
ultrasound and fluoroscopy to aid in stone localization,
many commercially available lithotripters now use a modular design in which the fluoroscopy unit is not attached to
the lithotripter reducing storage space as well as allowing
use of the fluoroscopy unit for other procedures.
MECHANISMS OF STONE FRAGMENTATION
Due to recent advances made in the basic science of shock
wave lithotripsy, there is a better understanding of how
shock waves result in stone fragmentation. Both the positive and negative portions of the shock wave (Fig. 1) are
critical in stone fragmentation and also play a role in renal
tissue injury. The four mechanisms described for stone
comminution include compressive fracture, spallation,
cavitation, and dynamic fatigue. As the calculus develops
in vivo, it is formed by both crystallization of minerals as
well as organic matrix material. This combination forms an
inhomogeneous and imperfect material that has natural
defects. When the shock wave encounters this inhomogeneous structure, the force generated in the plane of the
shock wave places stress on these imperfections resulting
in compression-induced tensile cracking. This mechanism
is known as compressive fracture. Spallation, another
mechanism of shock wave induced stone fragmentation,
occurs when the shock wave encounters fluid behind the
LITHOTRIPSY
stone and part of the wave is reflected back onto the stone
placing tensile stress on the same imperfections.
Shock waves cause bubbles to form on the surface of the
stone. These bubbles grow during the negative or tensile
component of the shock wave. When the positive component
of the next wave passes, these bubbles violently collapse
releasing their energy against the stone surface as secondary
shock waves and/or microjets. This phenomenon, known as
cavitation, represents the third mechanism of stone fragmentation. Dynamic fatigue describes the sum of accumulated
damage to the stone that coalesce to result in stone fragmentation and eventually destruction of the stone (10,11).
CLINICAL RESULTS OF SHOCK WAVE LITHOTRIPSY

Because many experts continue to consider the Dornier
HM3 (the original electrohydraulic lithotripter, Dornier
MedTech America, Inc., Kennesaw, GA) the gold standard
in lithotripsy, the available clinical literature comparing the
Dornier HM3 (first generation) lithotripter to other commercially available lithotripters is summarized in Table 1. This
type of summary does not truly allow a comparison between
different lithotripters currently in use. Yet, this comparison
demonstrates that modifications to second and third generation lithotripters have traded better patient comfort for a
lessening of stone free rates. This current clinical data has
been one of the driving forces behind the modifications that
are currently being investigated to improve SWL.
Summarized results of five large early series on the
clinical efficacy of SWL using the unmodified Dornier
HM3 lithotripter demonstrate stone free rates for renal
pelvis (RP), upper calyx (UC), middle calyx (MC), and lower
calyx (LC) stones were 76% (4885), 69% (4682), 68% (52
76), and 59% (4273), respectively. Stone free rates in these
series were better for smaller stones (010 mm) and RP
stones with relatively poorer stone free rates for LC stones.
In comparison, results of five large published series on the
Siemens Lithostar, a lower power machine demonstrated
stone free rates for RP, UC, MC, and LC stones of 69% (55
80), 67% (4690), 63% (4382), and 60% (4673). A comparison of integrated stone free rates stratified by size in a
regression model of the HM3 and Lithostar found significantly greater stone free rates across all stone sizes for the
HM3 lithotripter (11). While most studies have evaluated
the efficacy of SWL for adults, stone free rates with the HM3
are similar for children (21). The limited number of comparative studies of newer machines and the explosion in the
number of commercially available lithotripters makes it
difficult to assess their clinical efficacy.
While the HM3 has been shown to produce excellent
stone free rates for renal calculi, there continues to be
debate on the clinical efficacy of SWL for ureteral stones.
The main problem is that stones in the ureter are more
difficult to locate and therefore more difficult to target with
the shock wave. However, several studies have demonstrated stone free rates close to 100% for the treatment
of proximal ureteral stones with SWL (22). However, stone
free rates appear to decline to 70% for mid-ureteral stones
for many lithotripters (23). Treatment of distal ureteral
stones with SWL typically involves a prone or a modified
261
sitting position to allow shock wave targeting of the stone

below the pelvic brim. Stone free rates of distal ureteral
stones with the HM-3 lithotripter have been as high as
8596% (24). Endoscopic methods (ureteroscopy) can also
be employed for ureteral stones, especially those located in
the distal ureter with equivalent or better stone free rates.
While many of the lithotripters sighted in Table 1 are no
longer in clinical use or have been updated, these studies
clearly demonstrated several key points. The HM3 continues to provide equivalent and likely superior stone free
rates when compared to other lithotripters in studies.
While most commercially available lithotripters provide
acceptable stone free rates for stones located within the
kidney, these stones often require more treatment sessions
and adjunctive procedures to achieve the stone free rates of
the HM3 device. Additionally, success rates with all lithotripters declines the further the stone progresses down the
ureter and poses positioning challenges with alternative
methods indicated to remove ureteral stones.
TISSUE INJURY WITH CLINICAL LITHOTRIPSY

Clinical experience treating patients with SWL has demonstrated that while SWL is generally safe, shock waves can
cause acute and chronic renal injury. This concept has been
confirmed by multiple animal studies and a few human
clinical studies (11). Originally, shear stress to the tissue
was believed to be the main cause of this renal injury.
However, recent studies have sugggested that SWL
induced renal injury is a vascular event induced by vasoconstriction or cavitation-induced injury to the microvasculature (25). Additionally, this initial tissue damage may
promote further renal injury via the effects of free radical
production (26). Acute damage to the kidney appears to be
mainly a vascular insult.
While clinicians have long recognized the acute effects
of SWL, most have believed that there was no long-term
sequela to shock wave treatment. The > 25 years of clinical
experience with SWL serves as a testament to its safety
and effectiveness. However, several chronic problems may
develop as a consequence of SWL. Table 2 summarizes the
acute and chronic effects of SWL. Perhaps the most serious
long-term problem of SWL is the increased risk of hypertension. A review of 14 studies on the impact of SWL on blood
pressure demonstrated that when stratified according to the
number of shock waves administered, higher doses of shock
waves seem to correlate with a greater likelihood of increased
rates of new-onset hypertension or changes in diastolic blood
pressure (11). The impact of hypertension on cardiovascular
disease including the risk of myocardial infarction, stroke,
and renovascular disease make this a serious long-term effect
that needs further investigation.
Three mechanisms of SWL induced tissue injury have
been reported: cavitation, vasoconstriction, and free radical induced tissue injury. Investigators have demonstrated
in vitro that cavitation bubble expansion can rupture
artificial blood vessels (5). Other investigators have shown
in animal models that cavitation takes place in kidneys
exposed to shock waves (27). While cavitation bubbles that
form on the stone surface contribute to stone fragmentation,
Table 1. Literature Comparison of Lithotripters to Gold Standard Dornier HM3

Stone
Location
Number of
Patients
Stone Free Rates

(SFR), %
Wolf Piezolith
Kidney
HM3: 334
Wolf: 378
HM3: 75Wolf: 72
HM3: 15.5
Wolf: 45
13
EDAP LT01
Kidney
HM3: 500
EDAP: 500
HM3:77.2-90.4
EDAP: 42.5-87.5
> with EDAP
14
MFL 5000
Kidney
198 total
HM3: 80MFL: 56
15
Wolf Piezolith 2300
Ureter
70 total
HM3: 74Wolf: 76.6
16
Siemens Lithostar
Kidney
HM3: 91Siemens:85 HM3: 91Siemens: 65
17
EDAP LT01
Sonolith 2000
Kidney
HM3: 70EDAP:
and ureter
113Sono: 104
18
19
Kidney and
Siemens Lithostar,
ureter
Dornier HM4,
Wolf Piezolith 2300,
Direx Tripter X-L,
Breakstone
Medstone STS
Kidney
20
Lithotron
11
Siemens Lithostar
Study Type
Reference
Prospective
12
HM3 Compared to:
Auxiliary
Procedures
HM3: 4
Siemens: 13
262
Retrospective
Kidney
HM3: 79EDAP:
82Sono: 79
Retreatment
Rate, %
HM3: 12EDAP:
13Sono: 9
HM3: 4EDAP:
42Sono: 26
Multicenter
comparable between
2nd generation
HM3: 5698Med:
8166
HM3: 70Med: 81.5
HM3: 3.1Med: 5.5 HM3: 4.4

Med: 5.2
38 matched pairs
HM3: 79Lithotron:
58
> for Lithotron
Meta-analysis
HM3: 59-76
Siemens: 60-69
Comment
Wolf required more
retreatment, more shocks,
treatment rates
decreased dramatically
for ureteral stones
with Wolf
More sessions, increased
shocks required
with EDAP
Increased subcapsular
hematoma and longer
treatment times
with MFL
Comparable 3 month
SFR but used plain
radiographs for
comparison
SFR comparable at
3 months, Increased
tissue injury with
HM3 by urinary
enzymes
SFR at 3 months
comparable with
HM3 and EDAP, lower
for Sonolith, lower
retreatment with HM3
All were deemed inferior
to HM3 in terms of stone
free rates
Slightly better retreatment

and need for auxiliary
procedures with HM3
> for Lithotron HM3 superior to Lithotron
using matched pair
analysis
Using regression model,
SFR better with
HM3 across all
stone sizes
LITHOTRIPSY
263
Table 2. Acute and Chronic Injury with SWL
Changes to the Lithotripter
Acute
Chronic
Renal edema (swelling)
Hypertension(elevated
blood pressure)
Decreased renal function
Accelerated stone formation
(in animal models)
Renal scar formation
There are two major mechanical modifications that can

improve stone comminution, based on our current understanding of acoustic physics. One is to enhance the compressive component of the shock wave. The original HM3 relies on
a high energy output and thus the compressive component to
achieve stone fragmentation. The downside of this effect, from
clinical experience, is patient discomfort and potential renal
tissue injury. Alternatively, one can improve stone comminution by altering the tensile component of the shock wave and
thus better control cavitation. Below, several ways are
describe in which investigators are modifying the shock wave
to improve comminution, with decreased renal tissue injury.
Several investigators have modified lithotripters to alter
the negative portion of the shock wave that is responsible for
cavitational-induced stone fragmentation and tissue injury.
In one study, a reflector insert is placed over the original
reflector of an electrohydraulic lithotripter to create a second
shock wave that arrives behind the original shockwave, thus
partially canceling out the negative component of the shock
wave. These investigators found that this modification
reduced phantom blood vessel rupture, while preserving
stone fragmentation in vitro (29). Similarly, an acoustic diode
(AD) placed over the original reflector, has the same impact
as this modified reflector (30).
However, because reducing the tensile component of the
shock wave weakens that collapse of bubbles at the stone
surface, two groups have designed piezoelectric inserts into
an electrohydraulic lithotripter that send small, focused
shock waves at the time of bubble collapse near the stone
surface thus, intensifying the collapse of the cavitation bubble
without injuring surrounding tissue (29).
Another way in which investigators have modified the
shock wave is by delivering shock waves from two lithotripters to the same focal point, F2. Dual pulse lithotripsy
has been evaluated by several investigators both in vitro,
animal models and in clinical trials. Several investigators
have demonstrated in an in vitro model that the cavitation
effect became more localized and intense with the use of two
reflectors. Also, the volume and rate of stone disintegration
increased with the use of the two reflectors, with production
of fine (< 2 mm) fragments (31). In both animal models and
clinical studies, dual pulse lithotripsy has been shown to
improve stone fragmentation with reduced tissue injury.
Hematuria (blood in urine)

Subcapsular hematoma
Decreased renal
blood flow
Altered renal
function:
Impaired urine
concentration
Impaired control of
electrolytes
their formation in other locations (tissue, blood vessel lumen)

is an unwanted end product that results in tissue injury.
Recent investigations have elucidated yet another
potential mechanism of renal injury secondary to high
energy shock waves. Evidence suggests that SWL exerts
an acute change in renal hemodynamics (i.e., vasoconstriction) that occurs away from the volume targeted at F2, as
measured by a transient reduction in both glomerular
filtration rate (GFR) and renal plasma flow (RPF) (28).
Prolonged vasoconstriction can result in tissue ischemia
and permanent renal damage.
Vasoconstriction and cavitation both appear to injure the
renal microvasculature. However, as the vasoconstriction
induced by SWL abates, reperfusion of the injured tissue
might also result in further tissue injury by the release of free
radicals. These oxidants produced by the normal processes of
cellular metabolism and cellular injury cannot be cleared and
injure the cell membrane destroying cells. Free radical formation has been demonstrated in animal models (26).
It appears that the entire treated kidney is at risk of
renal damage from SWL-induced direct vascular and distant vasoconstrictive injury, both resulting in free radical
formation. Although previous studies have suggested that
the hemodynamic effects are transient in nature in normally functioning kidneys, patients with baseline renal
dysfunction may be at significant risk for permanent renal
damage (28). Patients of concern may be pediatric patients,
patients undergoing multiple SWL treatments to the same
kidney, patients with solitary kidneys, vascular insufficiency, glomerulosclerosis, glomerulonephritis, or renal
tubular insult from other causes.
SHOCK WAVE LITHOTRIPSY ADVANCES
Based on a better understanding of cavitation in stone
fragmentation as well as the role of cavitation, vasoconstriction, and free radical formation in SWL-induced tissue
injury, several groups are investigating ways in which
SWL can be clinically more effective and safe. In general,
these advancements involve changes to the shock wave
itself, by modifying the lithotripter, alterations in treatment technique, improvements in stone fragmentation /
passage or the reduction in tissue injury through medical
adjuncts, and improved patient selection.
Modifications to Treatment Strategy

The original Dornier HM3 lithotripter rate was synchronized
with the patients electrocardiogram so that the shock rate
did not exceed the physiologically normal heart rate. Experience with newer lithotripters has revealed that ungating the
shock wave delivery rate results in few cardiac abnormalities.
As a result, there has been a trend to deliver more shock
waves in a shorter period of time. However, increasing doses
of shock wave energy at a higher rate may have the potential
to increase acute and chronic renal damage.
As a result, several investigators have evaluated ways
in which the treatment strategy can be modified to optimize
SWL treatment. One proposed strategy is altering the rate
of shock wave delivery. Several investigators have reported
that 60 shocks min 1 at higher intensities resulted in the
264
LITHOTRIPSY
most efficient stone fragmentation than 120 shocks min 1.

This has been confirmed in vitro, in animal models and also
in randomized clinical trials. These studies speculate that at
increased rates, more cavitation bubbles were formed in
both the fluid and tissue surrounding the stone that did
not dissipate between shocks. As a result, these bubbles
scattered the energy of subsequent shocks resulting in
decreased efficiency of stone fragmentation (32).
In order to acclimate patients to shock waves in clinical
treatment, lower energy settings are typically used and
gradually increased. A study investigating whether
increasing voltage (and thus increasing treatment dose)
impacted on stone fragmentation has been performed in
vitro and in animal models. Stones fragmented into smaller
pieces when they were exposed to increasing energy compared to decreasing energy. The authors speculate that the
low voltage shock waves primed the stone for fragmentation at higher voltages (33). In addition, animals exposed to
an increasing voltage had less tissue injury than those
kidneys exposed to a decreasing or stable dose of energy.
While this treatment strategy has not been tested clinically, it might be able to improve in vivo stone comminution
while decreasing renal parenchymal injury (34).
In the same vein, several studies have reported that
pretreating the opposite pole of a kidney with a low voltage
dose of shock waves (12 kV), prior to treating a stone in the
other pole of a kidney with a normal dosage of shock waves,
reduced renal injury when as little as 100 low voltage shocks
were delivered to the lower pole. It is believed that the low
voltage shock waves causes vasoconstriction which protects
the treated pole of the kidney from hemorrhagic injury (35).
Other SWL treatment modifications being tested include
aligning the shock wave in front of the stone in order to
augment cavitation activity at the stone surface or apply
overpressure in order to force cavitation bubble collapse.
While these techniques have only been investigated in vitro,
these alterations in shock wave delivery, as well as the
previous treatment strategies demonstrate how an improved
understanding of the mechanisms of SWL can enhance stone
comminution and potential reduce renal tissue injury (36,37).
Adjuncts to Improve SWL Safety and Efficacy
Antioxidants. A number of studies have investigated
the role of antioxidants in protecting against free radical
injury to renal parenchyma (38). Table 3 summarizes the
results of various in vitro and in vivo studies on the use of
antioxidants to protect against SWL-induced renal injury
due to free radicals. While these studies are intriguing,
further clinical trials will be needed to evaluate potential
antioxidants for use in patients undergoing SWL.
Improving Stone Fragmentation. Another potential way
to improve stone fragmentation is to alter the stones
susceptibility to shock wave energy. One group has demonstrated that after medically pretreating stone in an in vitro
environment, one could achieve improved stone fragmentation. These data suggest that by altering the chemical
environment of the fluid surrounding the stones it is
possible to increase the fragility of renal calculi in vitro
(49). Further studies are warranted to see if calculi can be
Table 3. Investigated Antioxidants Providing Protection

against SWL-Induced Free Radical Injury
Reference
Study Type
39
In vitro
40
41
42
43
26
44
In vitro
In vitro, animal
Animal
Animal
Animal
Animal
45
46,47
48
Human
Human
Human
Antioxidant
nifedipine, verapamil,
diltiazem
Vitamin E, citrate
Selenium
Verapamil
Verapamil
Allopurinol
Astragulus membranaceus,
verapamil
Antioxidant vitamin
Verapamil, nifedipine
Mannitol
clinically modified, prior to SWL therapy, in the hopes of

enhanced stone fragmentation.
Improving Stone Expulsion. Several reports have demonstrated that calcium channel blockers, steroids, and alpha
blockers all may improve spontaneous passage of ureteral
stones. (ref) Compared to a control group, investigators found
improved stone clearance and shorter time to stone free in
patients treated with nifedipine or tamsulosin following SWL
compared to the control group. Additionally, retreatment rates
were lower (31%) for the medical treatment group compared to
the control group (51%). While expulsive therapy appears to
offer improved outcomes following SWL for ureteral stones,
confirmation with a randomized controlled study is needed
(50). Another intriguing report from the same group involves
the use of Phyllanthus niruri (Uriston) to improve stone
clearance of renal stones following SWL. This medication is
derived from a plant used in Brazilian folk medicine that has
been used to treat nephrolithiasis. Again, stone free rates were
improved with the administration of Uriston following SWL
compared to a control group with apparently the greatest
effect seen in those patients treated with lower pole stones (51).
Improving Stone/Patient Selection for SWL. Another way
to enhance the efficacy of SWL is improve patient selection.
Advances in computed tomography (CT) have allowed better
determination of internal stone architecture (52). As a consequence, a few studies have demonstrated that determining
the Hounsefield units (i.e., density unit of material on CT) of
renal stones on pretreatment, noncontrasted CT could predict stone free rates of patients treated with SWL (53).
Current micro-CT and newer multidetector CT scanners
have the potential to identify stone composition based on
CT attenuation. Therefore, stone compositions that are SWL
resistant, such as calcium oxalate monohydrate or cystine
stones, can be identified and those patients can be treated
with endoscopic modalities, therebye avoiding additional
procedures in these patients (54). Clinical trials utilizing this
concept will need to be performed.
Other factors, such as the distance of the stone from the
skin, weight of the patient, and other imaging modalities,
are being investigated to help determine who is likely to
benefit the most from SWL and which patients should be
treated initially with other modalities.
LITHOTRIPSY
CONCLUSIONS
Shock wave lithotripsy has revolutionized the way in which
urologists manage urinary calculi. Patients can now be
treated with minimal discomfort using an outpatient procedure. While all lithotripters rely on the same fundamental principles of acoustic physics, second and third
generation lithotripters appear to have traded patient
comfort and operator convenience for reduced stone free
rates, as compared to the original HM3 lithotripter. In
addition, mounting evidence demonstrates that SWL has
both acute and chronic impact on renal function and blood
pressure as a result of renal scarring.
Basic science research has provided insight into how
SWL results in stone comminution as well as renal tissue
injury. While the compressive component of the shock
wave causes stone comminution, it is apparent that the
tensile component plays a critical role in creating passable
stone fragments. These same forces cause tissue injury by
damaging the microvasculature of the kidney. This knowledge has resulted in several novel modifications to
improve both stone free rates as well as SWL safety.
Mechanical modifications to lithotripsy have focused on
controlling cavitation. Preventing bubble expansion in
blood vessels while intensifying bubble collapse near the
stone surface has been demonstrated to achieve improved
stone comminution with decreased tissue injury in vitro
and in animal models. Many of these designs could be
adapted to conventional lithotripters. Modification of
treatment techniques have also stemmed from our better
understanding of SWL. Slowing treatment rates may limit
the number of cavitation bubbles that can interfere with
the following shock wave. Voltage stepping and alternative-site pretreatment with low dose shock waves, may
cause global renal vasoconstriction that prevents cavitational injury to small vessels during treatment. In addition, our understanding that free radicals may be the end
culprit in parenchymal damage has suggested that pretreatment with antioxidants may prevent SWL-induced
renal injury. Finally, improved CT imaging may allow us
to predict which stones and patients are best suited for
SWL versus endoscopic stone removal. Further advances
will continue to make SWL a major weapon in the war
against stone disease for years to come.
BIBLIOGRAPHY
1. Lokhandwalla M, Sturtevant B. Fracture mechanics model of
stone comminution in ESWL and implications for tissue
damage. Phys Med Biol 2000;45:19231940.
2. Coleman AJ, Saunders JE, Crum LA, Dyson M. Acoustic
cavitation generated by an extracorporeal shockwave lithotripter. Ultrasound Med Biol 1987;13:6976.
3. Zhu S, Cocks FH, Preminger GM, Zhong P. The role of stress
waves and cavitation in stone comminution in shock wave
lithotripsy. Ultrasound Med Biol 2002;28:661671.
4. Howard D, Sturtevant B. In vitro study of the mechanical
effects of shock-wave lithotripsy. Ultrasound Med Biol
1997;23:11071122.
5. Zhong P, Zhou Y, Zhu S. Dynamics of bubble oscillation in
constrained media and mechanisms of vessel rupture in SWL.
265
6. Chaussy C, et al. First clinical experience with extracorporeally induced destruction of kidney stones by shock waves. J
Urol 1982;127:417420.
7. Preminger GM. Sonographic piezoelectric lithotripsy: More
bang for your buck. J Endourol 1989;3:321327.
8. Abernathy BB, et al. Evaluation of residual stone fragments
following lithotripsy: Sonography versus KUB. In: Lingeman
JE, Newman DM., editors. Shock Wave Lithotripsy II.
New York: Plenum Press; 1989. pp 247254.
9. Cartledge JJ, Cross WR, Lloyd SN, Joyce AD. The efficacy of a
range of contact media as coupling agents in extracorporeal
shockwave lithotripsy. BJU Int 2001;88:321324.
10. Eisenmenger W. The mechanisms of stone fragmentation in
ESWL. Ultrasound Med Biol 2001;27:683693.
11. Lingeman JE, Lifshitz DA, Evan AP. Surgical Management of
Urinary Lithiasis. In: Walsh PC, Retik AB, Vaughan ED, Jr.,
Wein AJ., editors. Campbells Urology. Philadelphia: Saunders; 2002. p 33613451.
12. Rassweiler J, et al. Wolf Piezolith 2200 versus the modified
Dornier HM3. Efficacy and range of indications. Eur Urol
1989;16:16.
13. Sofras F, et al. Extracorporeal shockwave lithotripsy or extracorporeal piezoelectric lithotripsy? Comparison of costs and
results. Br J Urol 1991;68:1517.
14. Chan SL, et al. A prospective trial comparing the efficacy and
complications of the modified Dornier HM3 and MFL 5000 lithotriptors for solitary renal calculi. J Urol 1995;153:17941797.
15. Francesca F, et al. Ureteral lithiasis: In situ piezoelectric
versus in situ spark gap lithotripsy. A randomized study. Arch
Esp Urol 1995;48:760763.
16. Graber SF, Danuser H, Hochreiter WW, Studer UE. A prospective randomized trial comparing 2 lithotriptors for stone disintegration and induced renal trauma. J Urol 2003;169: 5457.
17. Tan EC, Tung KH, Foo KT. Comparative studies of extracorporeal shock wave lithotripsy by Dornier HM3, EDAP LT 01
and Sonolith 2000 devices. J Urol 1991;146:294297.
18. Bierkens AF, et al. Efficacy of second generation lithotriptors:
A multicenter comparative study of 2,206 extracorporeal shock
wave lithotripsy treatments with the Siemens Lithostar, Dornier
HM4, Wolf Piezolith 2300, Direx Tripter X-1 and Breakstone
lithotriptors. J Urol 1992;148:10521056. Discussion 10561057.
19. Cass AS. Comparison of first generation (Dornier HM3) and
second generation (Medstone STS) lithotriptors: Treatment
results with 13,864 renal and ureteral calculi. J Urol
1995;153:588592.
20. Portis AJ, et al. Matched pair analysis of shock wave lithotripsy effectiveness for comparison of lithotriptors. J Urol
2003;169:5862.
21. Cass AS. Comparison of first-generation (Dornier HM3) and
second-generation (Medstone STS) lithotripters: Treatment
results with 145 renal and ureteral calculi in children. J
Endourol 1996;10:493499.
22. Robert M, ACh S, Lanfrey P, Guiter J, Navratil H. Piezoelectric shockwave lithotripsy of urinary calculi: Comparative
study of stone depth in kidney and ureter treatments. J
Endourol 1999;13:699703.
23. Marguet CG, Springhart WP, Auge BK, Preminger GM.
Advances in the surgical management of nephrolithiasis.
Minerva Urol Nefrol 2004;56:3348.
24. Rodrigues Netto Junior N, Lemos GC, Claro JF. In situ
extracorporeal shock wave lithotripsy for ureteral calculi. J
Urol 1990;144:253254.
25. Evan AP, et al. Shock wave lithotripsy-induced renal injury.
Am J Kidney Dis 1991;17:445450.
26. Munver R, et al. In vivo assessment of free radical activity
during shock wave lithotripsy using a microdialysis system: The
renoprotective action of allopurinol. J Urol 2002;167:327334.
266
27. Evan AP, et al. In vivo detection of cavitation in parenchyma of

the pig kidney during shock wave lithotripsy. American Urological Association Annual Meeting. Orlando (FL): 2002. p 1500.
28. Willis LR, et al. Effects of SWL on glomerular filtration rate
and renal plasma flow in uninephrectomized minipigs. J
Endourol 1997;11:2732.
29. Zhou Y, Cocks FH, Preminger GM, Zhong P. Innovations in
shock wave lithotripsy technology: updates in experimental
studies. J Urol 2004;172:18921898.
30. Zhu S, et al. Reduction of tissue injury in shock-wave lithotripsy by
using an acoustic diode. Ultrasound Med Biol 2004; 30:675682.
31. Sheir KZ, et al. Evaluation of synchronous twin pulse technique
for shock wave lithotripsy: determination of optimal parameters
for in vitro stone fragmentation. J Urol 2003; 170:21902194.
32. Paterson RF, et al. Stone fragmentation during shock wave
lithotripsy is improved by slowing the shock wave rate: studies
with a new animal model. J Urol 2002;168:22112215.
33. McAteer JA, et al. Voltage-Stepping During SWL Influences
Stone Breakage Independent of Total Energy Delivered:
In vitro studies with model stones. American Urological Association Annual Meeting. Chicago: 2003. p 1825.
34. Maloney M, et al. Treatment strategy improves the in vivo
stone comminution efficiency and reduces renal tissue injury
during shock wave lithotripsy. American Urological Association Annual Meeting. San Antonio (TX): 2005. p 1108.
35. Willis LR, et al. Same-pole application of low- and high-energy
shock waves protects kidney from swl-induced tissue injury.
American Urological Association Annual Meeting. San
Francisco: 2004. p 1114.
36. Sapozhnikov OA, et al. Effect of overpressure and pulse repetition frequency on cavitation in shock wave lithotripsy. J
Acoust Soc Am 2002;112:11831195.
37. Sokolov DL, et al. Prefocal alignment improves stone comminution in shockwave lithotripsy. J Endourol 2002;16:709715.
38. Preminger GM. Review: in vivo effects of extracorporeal shock
wave lithotripsy: animal studies. J Endourol 1993;7: 375378.
39. Jan CR, Chen WC, Wu SN, Tseng CJ. Nifedipine, verapamil
and diltiazem block shock-wave-induced rises in cytosolic
calcium in MDCK cells. Chin J Physiol 1998;41:181188.
40. Delvecchio F, et al. Citrate and Vitamin E Blunt the SWL
Induced Free radical surge in an in-vitro MDCK cell culture
model. American Urological Association Annual Meeting. San
Francisco: 2004. p 1120.
41. Strohmaier WL, Lahme S, Bichler KH. Amelioration of high
energy shock wave induced renal tubular injury by seleniuman in vivo study in rats. American Urological Association
Annual Meeting. Anaheim (CA): 2004. p 1529.
42. Yaman O, et al. Protective effect of verapamil on renal tissue
during shockwave application in rabbit model. J Endourol
1996;10:329333.
43. Willis LR, et al. Effects of extracorporeal shock wave lithotripsy to one kidney on bilateral glomerular filtration rate and
PAH clearance in minipigs. J Urol 1996;156:15021506.
44. Sheng BW, et al. Astragalus membranaceus reduces free
radical-mediated injury to renal tubules in rabbits receiving
high-energy shock waves. Chin Med J (Engl) 2005;118:4349.
45. Kehinde EO, et al. The effects of antioxidants on renal damage
occuring during treatment of renal calculi by lithotripsy.
American Urological Association Annual Meeting. San Antonio (TX): 2005. p 1698.
46. Strohmaier WL, et al. Protective effect of verapamil on shock
wave induced renal tubular dysfunction. J Urol 1993;150:2729.
47. Strohmaier WL, et al. Limitation of shock-wave-induced renal
tubular dysfunction by nifedipine. Eur Urol 1994;25:99104.
48. Ogiste JS, et al. The role of mannitol in alleviating renal injury
during extracorporeal shock wave lithotripsy. J Urol 2003;169:
875877.
49. Heimbach D, et al. The use of chemical treatments for improved

comminution of artificial stones. J Urol 2004;171: 17971801.
50. Micali S, et al. Efficacy of expulsive medical therapy using
nifedipine or tamsulosin after shock wave lithotripsy of ureteral stones. American Urological Association Annual Meeting.
San Antonio (TX): 2005. p 1680.
51. Antonio C, et al. May Phyllanthus niruri (Uriston) affect the
efficacy of ESWL on renal stnoes? A prospective, randomised
short term study. American Urological Association Annual
Meeting. San Antonio (TX): 2005. p 1696.
52. Zarse CA, et al. Nondestructive analysis of urinary calculi
using micro computed tomography. BMC Urol 2004;4:15.
53. Saw KC, et al. Calcium stone fragility is predicted by helical
CT attenuation values. J Endourol 2000;14:471474.
54. Williams JC, et al. Progress in the use of helical CT for imaging
urinary calculi. J Endourol 2004;18:937941.
See also MINIMALLY
INVASIVE SURGERY; ULTRASONIC IMAGING.
PAUL J. GAGLIO
Columbia University College
of Physicians and Surgeons
New York, New York
INTRODUCTION
From a conceptual perspective, liver transplantation involves
the replacement of a diseased or injured liver with a new
organ. Historically, liver transplantation has emerged from
an experimental procedure deemed heroic therapy for
patients not expected to survive, to the treatment of choice
with anticipated excellent long-term outcomes for patients
with end stage liver disease. This article will outline the
history of and indications for liver transplantation, delineate
short- and long-term complications associated with the procedure, and discuss the role of immunosuppressive therapy,
intrinsic to the long-term success of the procedure.
HISTORY
Historically, the most significant and persistent impediment to liver transplantation has been the availability of
suitable organs. Up until the early 1960s, death was
defined as cessation of circulation, and thus, donation from
deceased donors was thought to be both impractical and
impossible, as organs harvested from pulseless, nonperfusing donors would not function when transplanted, due to
massive cellular injury. The concept of brain death and
ability to harvest organs from individuals defined as such
first occurred at Massachusetts General Hospital in the
early 1960s, when a liver was harvested from a patient
whose heart was beating despite central nervous system
failure. This seminal event led to the development of a new
concept; death was defined when cessation of brain function
occurred, rather than the cessation of circulation. Thus, brain
dead donors with stable blood pressure and the absence of
comorbid disease could serve as potential organ donors.
Improvements in the ability to preserve and transport organs
dramatically increased organ availability, necessitating a
centralized system to facilitate procurement and allocation of

organs to individuals waiting for transplantation. This was
initially provided by SEOPF (the Southeast Organ Procurement Foundation), founded in 1968, from which UNOS (the
United Network for Organ Sharing) arose. At present, UNOS
operates the OPTN (Organ Procurement and Transplantation Network), providing a centralized agency that facilitates
recovery and transportation of organs for transplantation,
and appropriately matches donors and recipient.
LIVER TRANSPLANTATION: INITIAL RESULTS
The first reported liver transplantation occurred in 1955, in
the laboratory of Dr. Stuart Welch (1). In a dog model, an
auxiliary liver was transplanted into the abdominal cavity,
leaving the native liver in situ. Between 1956 and 1960,
various investigators initiated experiments in different animal
models whereby orthotopic liver transplantation was performed, achieved by removal of the native liver and implantation of a new liver in its place, requiring anastamoses of the
donor and recipient hepatic vein and artery, bile duct, and
portal vein (see Fig. 1). These initial attempts at liver transplantation refined the surgical procedure, however, graft dysfunction and death of the animals occurred quickly, due to
ineffective immunosuppression and eventual rejection of the
liver mediated by the animals immune system (2).
The first human liver transplants were performed by
Dr. Thomas Starzl in 1963, at the University of Colorado
(3). These early attempts at transplantation highlighted
the difficulties associated with extensive abdominal surgery in desperately ill patients, and were associated with
poor outcomes, largely due to technical difficulties and the
inability to effectively prevent rejection. Similar negative
experiences at other centers led to a worldwide moratorium
on liver transplantation. However, a major breakthrough
in the ability to prevent rejection and prolong the survival
of the transplanted liver occurred following the availability
of Cyclosporine in 1972 (described below). With continued
refinement of the surgical techniques required to perform
liver transplantation, combined with the ability to minimize
Suprahepatic
vena cava
Infrahepatic
vena cava
Portal vein
Bile duct
Hepatic artery
Figure 1. Schematic representation of an orthotopic liver
transplant.
267
organ rejection, posttransplant outcomes improved significantly. From 1963 to 1979, 170 patients underwent liver
transplantation at the University of Colorado; 56 survived
for 1 year, 25 for 1322 years, and several remain alive today
30 years after their surgery. Continued improvement in
posttransplantation outcomes were achieved, and thus, in
1983, the National Institutes of Health (NIH) established
that liver transplantation was no longer considered an
experimental procedure, rather, as definitive therapy for
appropriately selected patients with end-stage liver disease.
Additional advances in immunosuppression (reviewed below)
including the discovery of polyclonal and monoclonal antibodies to T-cells or their receptors, and other agents such as
Tacrolimus, Mycophenolate Mofetil, and Sirolimus have
further improved outcomes.
INDICATIONS FOR LIVER TRANSPLANTATION
Liver transplantation is an accepted therapeutic modality for
complications of chronic liver disease, or acute liver failure. In
general, liver transplantation is recommended when a
patient with end stage liver disease manifests signs and
symptoms of hepatic decompensation, not controlled by alternative therapeutic measures. This is evidenced by
1. Esophageal and/or gastric variceal bleeding, or
bleeding from portal hypertensive gastropathy.
2. Hepatic encephalopathy.
3. Spontaneous bacterial peritonitis.
4. Significant ascites.
5. Coagulopathy.
Patients with liver disease complicated by early stage
hepatocellular carcinoma (HCC), often defined as either a
single lesion <5 cm or not more than three lesions, each <3
cm are also considered candidates for liver transplantation
irrespective of evidence of concomitant hepatic decompensation (4).
If a patient meets these initial criteria, further requirements must be realized. It is generally accepted that liver
transplantation is indicated if the patient is not moribund
and the transplant is likely to prolong life with a > 50%
chance of 5-year survival. Furthermore, it is anticipated
that the transplant will restore the patient to a range of
physical and social function suitable for the activities of
daily living. Patients who are suitable candidates should
not have comorbid disease with involvement of another
major organ system, which would preclude surgery or
indicate a poor potential for rehabilitation.
Transplant candidates undergo a thorough psychological
assessment prior to liver transplantation. Adequate family
and social support must be demonstrated to ensure adherence to the difficult long-term medical regimen that will be
required posttransplant. In addition, if a history of substance
abuse is present, most transplantation programs require that
the patient complete at least 6 months of documented rehabilitation with displayed freedom from alcohol and/or drug
recidivism. Psycho-social assessment is usually performed
by several individuals, including a Psychiatrist or Psychologist and an experienced social worker. In addition, living
268
Table 1. Diseases Associated with Fulminant Hepatic

Failure
Viral Infection
Frequent
Rare
Metabolic
Toxin, Drugs
Drug Combos
Ischemic
Miscellaneous
Hepatitis A, B, D, E, Hepatitis Non A-G

Hepatitis C
Cytomegalovirus
Epstein Barr virus
Herpes simplex virus
Acute fatty liver of pregnancy
Reyes SX
Acetaminophen
Nsaids
CCL4
Isoniazid
Sodium valproate
Methyl DOPA
Tetracycline
Halothane
Amanita phalloides (mushroom poisoning)
Yellow Phosphorus
Herbal Medication
Acetaminophen and ETOH
Acetaminophen and barbiturates,
isoniazid trimethoprim, and
sulamethoxazole amoxicillin
and clavulinic acid
Hepatic artery thrombosis
Budd-Chiari Syndrome
Right ventricular failure, cardiac
tamponade shock
Hyperthermia
Hellp SX
donor liver transplantation (LDLT), discussed in greater

detail below, requires a detailed psychosocial assessment of
both recipient and potential donor. In most transplantation
centers, an independent donor advocate team consisting of a
social worker, internist, and surgeon who are independent of
the team evaluating the recipient performs the difficult task
of educating a potential donor regarding the risks and benefits of LDLT, assessing motivation to be a donor, and
determining if coercion is present.
Another indication for liver transplantation is fulminant hepatic failure, defined as hepatic encephalopathy
(confusion) arising in the setting of massive liver injury in a
patient without preexisting liver disease. This condition is
rapidly fatal unless recovery of hepatic function occurs
spontaneously, and thus, emergent liver transplantation
may be required. Conditions associated with fulminant
hepatic are listed in Table 1.
ETIOLOGY OF LIVER DISEASES REQUIRING LIVER
TRANSPLANTATION
Diseases associated with hepatic dysfunction in adults and
children are outlined in Tables 2 and 3, respectively. In general,
any disease process in adults or children that induces either
acute or chronic hepatocellular, biliary, or vascular injury may
necessitate liver transplantation. The indications for liver
transplantation in children are identical to those in adults, that
is, liver transplantation is indicated in the presence of progressive liver disease in patients who fail medical management.
Table 2. Indications for Liver Transplantation

Diseases Effecting Hepatic Parenchyma
Viral hepatitis with cirrhosis (Hepatitis B with or without
Delta Virus, Hepatitis C, Non A-E hepatitis)
Autoimmune hepatitis
Alcoholic cirrhosis
Metabolic disorders (Wilsons disease, hemochromatosis, alpha 1
Antitrypsin, Tyrosinemia, protoporphyria, Cystic fibrosis,
familial amyloidosis, Neiman-Pick disease)
Fulminant hepatic failure due to any cause
Drug induced liver disease
Diseases Effecting Biliary System
Primary and secondary biliary cirrhosis
Sclerosing cholangitis
Carolis disease
Relapsing cholangiohepatitis
Choledochal cysts with obstruction and bilary cirrhosis
Hepatic Neoplasia/Malignancies
Patients with nonmetastatic primary hepatocellular carcinoma,
with;
A single tumor not > 5 cm
No more than three lesions with the largest lesion < 3 cm
No thrombosis of the portal or hepatic vein
Hemangioendothelioma (confined to the liver)
Neuro endocrine tumors with hepatic involvement
Large hepatic Hemangioma
Miscellaneous Causes
Hepatic vein thrombosis (Budd-Chiari syndrome)
Portal vein thrombosis
Trauma
Many of the disease processes in adults that induce liver failure

are recapitulated in children. However, specific disease states
seen in children including metabolic diseases and congenital
biliary anomalies represent additional indications for liver
transplant. Moreover, liver transplantation is indicated in
infants and children if the transplant will prevent or attenuate
derangements in cognition, growth, and nutrition. Therefore,
children should be considered for liver transplantation when
there is evidence that hepatic decompensation is either unavoidable (based on knowledge of the history of the disease itself),
imminent, or has already occurred. The clinical scenarios that
determine when liver transplantation is required in children
can include one or more of the following:
1.
2.
3.
4.
5.
6.
7.
8.
9.
Intractable cholestasis.
Portal hypertension with or without variceal bleeding.
Multiple episodes of ascending cholangitis.
Failure of synthetic function (coagulopathy, low
serum albumin, low cholesterol).
Failure to thrive or achieve normal growth, and/or
the presence of cognitive impairment due to metabolic derangements, and malnutrition.
Intractable ascites.
Encephalopathy.
Unacceptable quality of life including failure to be
able to attend school, intractable pruritis.
Metabolic defects for which liver transplantation
will reverse life-threatening illness and/or prevent
irreversible central nervous system damage.
269
Table 3. Additional Indications for Liver Transplantation in Infants and Children

Cholestatic Liver Disease
Obstructive: Biliary Atresia (most common indication for liver transplantation in children)
Intrahepatic: Alagilles Syndrome, Bylers disease, familial cholestatic symptoms
Other
Congenital hepatic fibrosis
Metabolic Diseases
Disease
Defect
Inheritance
Alpha 1 antitrypsin
Wilsonss Disease
Tyrosinemia
Urea cycle defects
Galactosemia
Glycogen storage Diseases
Type 1A
Type IV
Familial
hypercholesteroloemia
Gauchers Disease
Nieman-Pick disease
Crigler-Najjar type 1
Cystic fibrosis
Hyperoxaluria type 1
Neonatal Fe storage
Hemophilia A and B
Comments
Decreased serum A1AT

Decreased Ceruloplasmin
Codominant
May reverse both liver and lung disease
Autosomal
Recessive (AR)
Fumarylacetoacetate hyrolase
AR
Transplant in fulminant liver failure,
or to prevent hepatic neoplasia
Example: ornithine
x-linked
Prevent CNS injury
transcarbamylase
dominant
Arginosuccinate synthetase
AR
Galactose phosphate uridyl transferase AR
Prevent development of cirrhosis and Hepatoma
Glucose 6 phosphatase
AR
Consider transplant if dietary management not
successful
Brancher enzyme
Type 2 A-LDL receptor deficiency
AR
Avoids ASHD
Glucocerebrosidase
Sphingomyelinase
Uridine diphosphate glucoronly
transferase
Chloride ion transfer gene
Alanine glyoxalate aminotransferase
Unknown
Factor VIII/IX
AR
AR
AR
May need combined liver/bone marrow t-plant
Disorders of bile acid

Unknown
synthesis (Bylers disease)
10. Life threatening complications of stable liver disease (e.g., hepatopulmonary syndrome).
AR
AR
Varies
x-linked
Varies
prevents fatal Kernicterus

May need combined liver/lung transplant
Usually requires combined liver/kidney
Transplant as infant
Transplant indication varies (?iron overload,
factor inhibitor present)
Transplant indicated if associated with end stage
liver disease
comes must be restrained by evidence that HCV recurrence

in HIVHCV coinfected patients may be problematic (5).
CONTRAINDICATIONS TO LIVER TRANSPLANTATION

(ADULTS AND CHILDREN)
RECIPIENT CHARACTERISTICS AND PRIORITIZATION FOR

TRANSPLANTATION
At present, absolute exclusion criteria for liver transplantation are evolving. In general, patients with advanced
cardiac or pulmonary disease, severe pulmonary hypertension, active substance abuse, coma with evidence of
irreversible central nervous system injury, sepsis, or
uncorrectable congenital abnormalities that are severe
and life threatening are not transplant candidates. In
addition, individuals with evidence of extrahepatic malignancy do not meet criteria for transplantation, unless the
patient meets standard oncologic criteria for cure. Relative exclusion criteria include renal insufficiency when
renal transplantation is not feasible, prolonged respiratory
failure requiring > 50% oxygen, advanced malnutrition,
primary biliary malignancy, inability to understand the
risk/benefits of the procedure, and inability to comply with
medications and conform to follow-up regimens. Recent
data indicates successful outcomes in HIV infected patients
who undergo liver transplantation, a population formerly
considered noncandidates for the procedure. However,
initial enthusiasm regarding successful transplantation out-
Given the relatively stable number of available donor

organs in the setting of a rapidly expanding pool of potential recipients, the timing of transplantation is critical.
Liver transplantation in a stable patient who is anticipated
to do well for many years while waiting for an available
organ may not be appropriate, while liver transplantation
in a moribund patient with a low probability of posttransplantation survival is similarly inappropriate. Prior to
1997, prioritization for liver transplantation was based
on the location where patients received their care (i.e.,
home, hospital, intensive care unit) and waiting time on
the transplant list. In 2002, several policies were instituted
by UNOS in an attempt to produce a more equitable organ
allocation scheme. Waiting time and whether the patient
was hospitalized were eliminated as determinants of prioritization of organ allocation. The MELD score (Model for
End Stage Liver Disease) a logarithmic numerical score
based on the candidates renal function (creatinine), total
bilirubin, and INR (international normalized ratio for prothrombin time) has been shown to be the best predictor of
270
mortality among cirrhotic patients, including those on the

transplant waiting list. It was therefore adopted by UNOS
as a mechanism to prioritize waiting list candidates.
MELD had been validated as a predictor of 3-month survival in diverse groups of patients with various etiologies
and manifestations of liver disease (6). Presently, a
patients position on the liver transplantation waiting list
is now determined by their MELD score; patients with
highest MELD scores are ranked highest on the list.
Prospective analysis of the impact of MELD indicates
improvement in both the rate of transplantation, pretransplantation mortality, and short-term posttransplantation
mortality rates (7). However, retrospective analysis has
suggested that posttransplantation survival may be
reduced in patients with very high pretransplantation
MELD score, particularly in Hepatitis C infected patients
(8). Conversely, MELD score effectively delineates when a
patient is too well for transplantation. A recent review
indicates that posttransplantation survival in patients
transplanted with a MELD score of <15 is lower than a
nontransplanted cohort with similar MELD score (9).
Thus, it is clear that careful recipient selection, with
attention to pressor and ventilatory requirements, need
for dialysis, age, and MELD score are important factors in
selecting appropriate candidates for liver transplantation.
LIVER TRANSPLANTATION: SOURCE OF ORGANS

At present, there are three potential types of organ donors
specific to liver transplantation, identified as deceased, living, or non-heart beating. Deceased donors (DD) comprise
the majority of liver donors. Either by self-identification
while living, or after discussion with next of kin when
donor brain death has been declared, individuals are
acknowledged as potential organ donors. Recent data from
UNOS indicate that 1- and 3-year patient survival in recipients of DD liver transplant is 81 and 71%, respectively
(10). However, despite efforts to maximize utilization of
organs acquired from DD including the use of older donors,
steatotic (fatty) livers, and livers infected with Hepatitis C
or B, a growing disparity exists between the number of
available livers and the number of individuals waiting
for transplantation. This critical shortage of organs has
resulted in both an increase in the waiting time for liver
transplantation and death rate among patients on the
waiting list. In response, the modalities of adult-to-child
and adult-to-adult LDLT have emerged as alternatives to
deceased donor liver transplantation (11,12). Adult-to-child
LDLT usually involves the removal of the left lateral segment of the liver (20% of hepatic mass) from an adult donor
for implantation into a child, while adult to adult living
donor liver transplantation requires that the larger, right
lobe of the liver (which accounts for 5060% of the hepatic
mass) be removed from the donor to ensure adequate hepatic mass in the recipient. Rapid regeneration of the liver
remnant in the donor and the partial allograft transplanted
into the recipient occurs, to the extent that appropriate liver
volume is restored within 12 months in both donor and
recipient following surgery. Since most pediatric LDLT
recipients are <2-years old, they receive a liver graft of
adequate or even excessive size, and thus liver insufficiency

due to the receipt of inadequate liver mass is rare. In
contradistinction, as the recipient of an adult-to-adult living
donor liver transplantation receives a graft that must over
time grow to an appropriate volume, selection of recipients
best able to tolerate transplantation of a partial graft is
necessary. In appropriately selected pediatric and adult
recipients, 1- and 3-year graft and patient survival in individuals who undergo LDLT is similar or superior to DD (10).
However, when comparing postoperative complications in
recipients of DD versus LDLT, recipients of LDLT have a
greater rate of biliary complications including bile leaks and
biliary strictures, which occur in 1532% of patients (13). In
addition, the small-for-size syndrome manifested as prolonged posttransplantation cholestasis with or without portal hypertension may occur following LDLT, if the graft is of
inadequate size (14). Fortunately, the majority of patients
who experience this syndrome recover without the requirement of retransplantation.
Recently, significant interest in the utilization of nonheart beating donors (donation after cardiac death,
DACD) as a potential modality to further increase the pool
of available organs has emerged. In contrast to DD who are
declared brain dead, DACD are critically ill patients who
are not brain dead, but have no expectation of recovery and
who based on their own prior wishes or families request are
removed from life support. Following cardiac arrest and
declaration of death, organs are harvested. There are two
types of DACD, controlled and uncontrolled. In the
controlled DACD (Maastricht category 3 death anticipated) the patient is removed from life support and death
occurs in the operating room. Once death has been
declared, organs deemed suitable for transplantation are
rapidly perfused with preservation solution and removed
surgically. The uncontrolled DACD (Maastricht category
1 and 2 death not anticipated) is declared dead after
cardiac arrest, rushed to the operating room, and organs
are harvested. Uncontrolled DACD are usually not utilized for liver transplantation due to the high rate of
primary nonfunction (defined below), usually due to prolonged ischemia of the graft. When utilizing controlled
DACD for transplantation, emerging data indicates that
recipient and graft survival are diminished when compared to deceased and living donor liver transplantation
with a higher incidence of primary nonfunction, biliary
injury, and requirement for retransplantation. However,
several centers have reported acceptable outcomes when
utilizing controlled DACD organs, particularly those without
significant ischemia in well-selected recipients (15).
Finally, domino transplantation is an option for
patients afflicted with familial amyloidotic polyneuropathy
(FAP). Familial amyloidotic polyneuropathy is a fatal disease caused by an abnormal amyloidogenic transthyretin
(TTR) variant generated by the liver. Liver transplantation
in these patients removes the source of the variant TTR
molecule, and represents the only known curative treatment. As no intrinsic liver disease exists in patients
affected by FAP, the liver explanted from a patient with
FAP may be transplanted into another patient, thus,
allowing domino transplantation. Survival in both recipients of FAP livers and transplanted FAP patients has
been reported to be excellent and comparable to survival

with OLT performed for other chronic liver disorders (16).
POSTTRANSPLANTATION MANAGEMENT
The complex nature of the surgical procedure utilized to both
explant (remove) the diseased, cirrhotic liver and implant
(transplant) the new allograft into the recipient make it
intuitive that the majority of the early complications following liver transplantation are technical and related to the
surgical procedure itself. However, following the first postoperative days, and as patients progress to the first month
posttransplantation and beyond, the nature and variety of
complications change. Early complications (within the first
2 months) and late complications (beyond 2 months) may
negatively affect patient and graft survival (Table 4). Complications specific to the surgical procedure and those that
directly affect the transplanted organ are discussed below.
EARLY COMPLICATIONS
Primary Nonfunction and Early Graft Dysfunction
A major threat to the newly transplanted liver is primary
graft nonfunction (PNF). This syndrome defined as acidosis, rising INR, progressive elevation in liver transaminases and creatinine, and decreases in mentation occurs
when the newly transplanted liver allograft fails to function normally. The mechanisms responsible for this phenomenon are complex, and relate to donor factors,
Table 4. Early and Late Complications Following Liver
Transplantation
Early
Graft Specific
Primary nonfunction
Early graft dysfunction
Hepatic and portal vein thrombosis
Preservation injuryBiliary complications: bile leak, biliary
stenosis
Acute cellular rejection
Other
Bacterial and fungal infection
CMV infection
Recurrent Hepatitis B and C
Late
Graft Specific
Chronic rejection
Recurrence of primary disease
Other
Hypertension
Hyperlipidemia
Diabetes
Obesity
Cardiac disease
Renal dysfunction
Fungal infection (Cryptococcus, Aspergillus)
CMV
Posttransplant lymphoproliferative disorder
Nonhepatic malignancy: i.e., skin cancer
271
inadequate preservation of the liver, prolonged ischemia,

extensive steatosis of the graft, hepatic artery thrombosis
(see below) or immune response to the implanted organ
(17). In the setting of PNF, a rapid assessment of hepatic
artery flow needs to occur, as immediate surgical repair of a
thrombosed hepatic artery may reverse PNF. In the
absence of hepatic artery thrombosis, emergent retransplantation is required for PNF.
In contrast to PNF, early graft dysfunction (EGD) is
manifested by an early rise in serum transaminases to values
> 20003000 IU/L, cholestasis with rising Bilirubin levels,
without marked coagulopathy or impairment in mental status and renal function. EGD may occur in the setting of
ischemic injury or steatosis in the graft, and typically occurs
within the first 2448 h after the transplant. Unlike PNF, the
manifestations of EGD usually improve with supportive care,
and emergency retransplantation is not necessary.
Hepatic Artery Thrombosis
A potentially devastating posttransplantation complication is hepatic artery thrombosis (HAT). Hepatic artery
thrombosis occurs more commonly in pediatric transplant
recipients compared to adults due to the technical difficulties associated with the anastomosis of smaller size
vessels. In HAT, the immediate postoperative period may
be associated with graft failure, elevation in serum liver
transaminases, bile leak, hepatic necrosis, and sepsis.
Since the blood supply to the biliary tree in the early
posttransplant period is principally from the hepatic
artery, HAT is frequently associated with irreversible
injury to the biliary tract (18). Thus, HAT in the first 7
days after liver transplantation is an indication for emergent artery repair or retransplantation.
Due to the potentially devastating consequences of HAT,
most transplant centers screen for this complication with
duplex-ultrasound (US) in the immediate posttransplant period. If duplex-US suggests HAT, angiography is usually
performed to confirm the diagnosis, and if present, surgical
revision of the hepatic artery is required. If surgical repair
cannot be achieved, liver retransplantation may be necessary.
PORTAL AND HEPATIC VEIN THROMBOSIS
Though less common than HAT, thrombosis of the portal
and/or hepatic veins in the immediate posttransplant period can also adversely affect patient and graft survival.
Acute Budd-Chiari syndrome due to hepatic vein or vena
cava thrombosis is associated with abdominal pain, peripheral edema, and the threat of graft failure, as hepatic
congestion in the newly transplanted liver is poorly tolerated. In this circumstance, emergency thrombectomy and
surgical revision is required. Acute portal vein occlusion
may be associated with exacerbation of preexisting portal
hypertension, associated with gastrointestinal bleeding
from porto-systemic collateral vessels such as esophageal
and gastric varices. Acute portal vein thrombosis is managed by surgical repair, while chronic portal vein thrombosis may be well tolerated. A potential alternative to
surgical repair for both hepatic and portal vein stenosis
or occlusion is thrombolysis and/or the placement of
272
endovascular stents by an experienced interventional radiologist (19).

ACUTE CELLULAR REJECTION
Rejection of any transplanted organ is a constant threat, as
immunologic recognition of the graft as foreign may be
associated with injury. However, compared to other
organs, liver allografts are relatively privileged immunologically, and thus, the incidence and consequences of
acute cellular rejection (ACR) are diminished when compared to other solid organs utilized for transplantation.
The reported incidence of ACR within the first posttransplant year is 3050%, in most cases, usually occurring
within the first 23 weeks postoperatively. The clinical
presentation is variable; ACR may be asymptomatic, or
associated with fever or abdominal pain. Laboratory findings include elevation or failure of normalization of serum
transaminases, usually in association with a rising alkaline phosphatase and/or bilirubin. The diagnosis of acute
liver graft rejection is confirmed by liver biopsy and examination of liver histology (20). Conventional histologic
criteria associated with ACR include the presence of periportal lymphocytic infiltrate, as well as bile duct and
hepatic vascular endothelial cell injury. Most cases of
ACR respond to treatment with intravenous bolus glucocorticoids. Approximately 10% of patients with ACR will
not improve with intravenous glucocorticoids, requiring
the administration of monoclonal or polyclonal anti-T cell
antibodies (reviewed below). Mild and moderate ACR may
also respond to either increasing the dose of the primary
immunosuppressive agent, or switching to an alternate
calcineurin inhibitor. This approach has been used with
increasing frequency, particularly in patients transplanted
for HCV and HBV due to concerns regarding the negative
impact of over-immunosuppression on viral recurrence.
BILIARY COMPLICATIONS
Bile leaks and strictures generally occur at the anastomosis of the donor and recipient bile ducts, recognized by a rise
in serum bilirubin and/or alkaline phosphatase or by the
presence of bile in surgical drains in the immediate posttransplantation period. The incidence of biliary complications is between 5 and 15% following deceased donor liver
transplantation. However, between 15 and 30% of patients
who undergo living donor liver transplantation develop
biliary complications, due to the complexity of the biliary
reconstruction required (21). In both deceased and living
donor recipients, the majority of bile leaks resolve spontaneously without the need for reoperation. As previously
stated, the biliary tree receives the vast majority of its
blood supply from the hepatic artery, and thus, the adequacy of hepatic artery blood flow needs to be evaluated in
the setting of any biliary injury. If spontaneous resolution
of the bile leak does not occur, endoscopic or radiologic
placement of a biliary stent across the biliary anastamoses
is often successful (22). In some cases, surgical exploration
and revision of the biliary anastamoses with a Roux-en-Y
choledochojejunostomy may be required.
Anastamotic biliary strictures require careful attention,

as if left untreated, cholangitis, graft dysfunction, and
eventually secondary biliary cirrhosis may occur. Techniques for management include dilatation and stenting via
biliary endoscopy or percutaneous transhepatic cholangiogram by an interventional radiologist. If these modalities
are unsuccessful, surgical revision of the biliary anastamosis with a Roux-en-Y choledochojejunostomy may be
required. In rare cases with diffuse stricturing, retransplantation may be necessary.
ISCHEMIC AND PRESERVATION INJURY
The newly transplanted liver is always subjected to some
degree of ischemic injury (23). Cold (or hypothermic) ischemia is unavoidable, as it occurs prior to transplantation
while the liver is cooled in preservation solution, awaiting
implantation. Warm (normothermic) ischemia occurs during the transplantation procedure itself, when hepatic
blood flow is interrupted to minimize blood loss during
transplantation, or when the formerly cooled liver is
subjected to body temperature during transplantation.
Cold ischemia is usually well tolerated, while in contrast,
warm ischemia often leads to death of hepatocytes, with
resultant elevation in serum transaminases, apoptosis and
centrilobular necrosis. In the setting of significant warm
ischemia, graft failure may result. Several investigators
have noted improvement in ischemic injury and enhanced
graft and patient outcomes by employing a technique
described as ischemic preconditioning defined as a brief
period of controlled ischemia followed by a short interval of
reperfusion before the actual surgical procedure (24). This
is accomplished during liver transplantation by transiently
interrupting hepatic inflow by placing a vascular clamp or
a loop around the portal triad (i.e., portal vein, hepatic
artery, and bile duct), rendering the whole organ ischemic
for 1015 min, after which the clamp is removed and the
liver is reperfused. This technique may be of particular
benefit in organs with significant steatosis.
Complications Beyond Two Months
Progress in the surgical techniques required to perform
transplantation, the treatment of postoperative complications and prevention of rejection have been associated with
significant improvements in short-term morbidity and mortality following transplantation. Coincident with improvements in short-term outcomes has been a rise in long-term
complications. These complications, including side effects of
chronic immunosuppression, neoplasia, and infections are
discussed in detail elsewhere. Long-term complications that
affect the transplanted liver are discussed below.
CHRONIC REJECTION
Chronic allograft rejection or vanishing bile duct syndrome is rare, but in contradistinction to acute cellular
rejection, a much more difficult to treat complication.
Diagnostic criteria for chronic rejection include bile duct
atrophy affecting the majority of bile ducts, with or without bile duct loss. Arterial and venous injury affecting the
large branches of the hepatic artery or portal vein (foamy

arteriopathy) may also be present (24). Risk factors for
chronic liver rejection include transplantation for primary sclerosing cholangitis, primary biliary cirrhosis,
HLA mismatch between donor and recipient, and cytomegalovirus infection. Chronic rejection is usually a harbinger
of poor outcomes, often resulting in the requirement for
retransplantation; altering immunosuppression is rarely
associated with improvement.
Recurrence of Primary Disease Following Liver
Transplantation
A major challenge to the liver transplant community is recurrence of the primary disease that caused the patients native
liver to fail. Diseases that do not recur following liver transplantation include congenital anatomic anomalies (e.g., biliary
atresia, polycystic liver disease, Carolis disease, Alagilles
syndrome, congenital hepatic fibrosis) and metabolic diseases
of the liver (e.g., Wilsons disease, alpha 1 antitrypsin deficiency). However, all other causes of liver disease including
primary biliary cirrhosis, primary sclerosing cholangitis, autoimmune hepatitis, nonalcoholic fatty liver disease, hemochromatosis and alcohol related liver disease have been reported to
recur after liver transplantation. In some cases, recurrent
disease may lead to significant liver injury with resultant
graft failure (2630). Disease processes most commonly associated with recurrence include viral hepatitis B (HBV) and C
(HCV). The recurrence of HBV is associated with uniformly
poor outcomes with graft failure and death. Fortunately,
recurrence of HBV after liver transplantation can be prevented by administering hepatitis B immune globulin (HBIG)
at the time of transplantation and at regular intervals thereafter, with or without the use of antiviral agents such as
Lamivudine and Adefovir. In contradisctinction to HBV,
HCV recurrence following liver transplantation remains a
significant source of morbidity and mortality, with negative
impact on post-transplantation outcomes. In patients with
active HCV replication prior to transplantation, reacquisition
of viremia following transplantation is universal, and histologic injury due to HCV occurs in up to 90% of patients
followed for 5 years (31). Although histologic injury in the
allograft due to HCV is exceedingly common, disease progression after the development of hepatitis is variable, with
some patients experiencing indolent disease and others
rapidly progressing to cirrhosis and liver failure. In patients
that develop HCV associated cirrhosis posttransplantation,
up to 42% will experience decompensation manifested as
ascites, encephalopathy, or hepatic hydrothorax, and <50%
of patients survive > 1 year after the development of decompensation (32). It is important to contrast the natural history of HCV before and after transplant; prospective and
retrospective data are emerging which indicate that the
progression of HCV following liver transplantation is
accelerated when compared to the nonimmunosuppressed
pretransplant patient population.
Whether HCV recurrence is more severe in recipients of
LDLT than in DD recipients is controversial. Although
several recent reports indicate that HCV recurrence may
be more problematic in recipients of LDLT when compared
to DD (33), particularly the cholestatic variant of HCV (34),
273
other authors have noted no differences in outcomes in

HCV infected patients who undergo LDLT when compared
to DD (35,36). At present, both the optimal timing for
transplant in HCV patients and the therapy for recurrent
HCV following liver transplantation are incompletely
described. Theoretically, eradication of HCV prior to liver
transplantation in patients with decompensated liver disease would be beneficial, although in practice, this strategy
has been marred by exacerbation of encephalopathy, infections, and other serious adverse events, particularly in
patients treated with high dose Interferon and ribavirin
(37). A novel approach including initiating therapy with
low dose interferon (including Pegylated interferon preparations) and ribavirin with slow escalation in dose may
be associated with improved tolerability and efficacy (38).
Following liver transplantation, both preemptive therapy
prior to the development of histologic injury and directed
therapy after the onset of liver injury have been attempted
with varying degrees of success. It is important to note,
however, that posttransplantation, tolerability of interferon preparations, and ribavirin is suboptimal. Significant
leucopenia and anemia are common, likely due to drug
induced bone marrow suppression and renal insufficiency
potentiating ribavirin induced hemolysis (39).
Immunosuppressive Medications
A cornerstone to posttransplantation management is the ability to prevent or attenuate immunologic rejection of
the transplanted graft, which when left untreated, can be
associated with graft failure. From a conceptual standpoint,
understanding how recognition of the newly engrafted liver as
foreign occurs, how to modulate immune mediated injury,
and at the same time prevent overimmunosuppression
are critical to achieve optimal post transplantation outcomes. The various immunosuppressive medications and
their mechanism of action currently utilized in liver transplant recipients are listed in Table 5. Unfortunately, all
immunosuppressive therapy is associated with undesired
effects, with the potential for additive effects when agents
are combined. In general, most transplant centers utilize
three agents to prevent allograft rejection in the immediate
posttransplant period, utilizing a combination of a calcineurin inhibitor such as Cyclosporine (CYA) or Tacrolimus
(TAC), a second agent such as Mycophenolate mofetil (MMF)
or Azathioprine (AZA), and a glucocorticoid such as Prednisone. As patients achieve adequate liver function and freedom from rejection beyond 6-months posttransplantation,
satisfactory immunosuppression can be achieved in many
patients with monotherapy, usually with a calcineurin inhibitor, although in patients who are at increased risk of
rejection such as those with autoimmune hepatitis, primary
biliary cirrhosis, or sclerosing cholangitis, long-term immunosuppression is achieved with a combination of a calcineurin
inhibitor with either low dose MMF or Prednisone (40).
Corticosteroids
Corticosteroids achieve their desired immunosuppressive
affects by the suppression of leukocyte, macrophage, and
cytotoxic T-cell activity, and diminution of the effect of
274
Table 5. Immunosuppressive Agents
Agent
Antilymphocyte
globulin
Antithymocyte
globulin
OKT3
Mechanism of Action
Side Effects
Depletes circulating lymphocytes
Flu-like symptoms
Anaphylaxsis
Lymphoproliferative disorders
Depletes circulating T cells
Flu-like symptoms
Anaphylaxsis
Lymphoproliferative disorders
Infections
Gastrointestinal distress
Pulmonary edema and
bronchospasm (rare)
Hypertension
Renal insufficiency
Neuropathy
Hyperlipidemia
Gingival hyperplasia
HirsutismInsulin resistance
Hypertension
Dyslipidemia
Glucose intolerance
Bone abnormalities
Peptic ulcers
Psychiatric disorders
Leukopenia
Anemia
Thrombocytopenia
Pancreatitis
Hypertension
Renal insufficiency
Insulin resistance
Neuropathy
Hyperlipidemia
Leukopenia
Anemia
Thrombocytopenia
GI side effects
Bone marrow suppression
Hyperlipidemia
Pneumonitis
Inhibits wound healing
Basiliximab
Daclizumab
IL-2 receptor blockade
Cyclosporine
Inactivates calcineurin, decreases

IL2 production, Inhibits
T-cell activation
Prednisone
Suppression of leukocyte, macrophage,

and cytotoxic T-cell activity
Decrease cytokines, prostoglandins,
and leukotrienes
Azathioprine
Inhibits adenosine and guanine production

Inhibits DNA and RNA synthesis in rapidly
proliferating T cells
Tacrolimus
Inactivates calcineurin, decreases IL2

production, Inhibits T-cell activation
Mycophenolate
mofetil
Inhibits of inosine monophosphate

dehydrogenase (IMPDH)
Prevents T- and B-cell proliferation
Sirolimus
inhibiting mTOR (target of Rapamycin)

Prevents T-cell replication.
cytokines, prostaglandins, and leukotrienes. However, hypertension, dyslipidemia, glucose intolerance, bone loss, peptic
ulcers and psychiatric disorders are often associated with
therapy. Therefore, a strategy to taper and discontinue glucocorticoids within the first 6 months1 year following transplantation while maintaining adequate levels of calcineurin
inhibitor is employed by many transplant centers. This tactic is
often altered in patients who undergo liver transplantation
secondary to an immunologic disorder such as autoimmune
hepatitis, primary biliary cirrhosis and sclerosing cholangitis
due to an enhanced risk of acute cellular rejection. In these
patients, either long-term use of corticosteroids with an attempt
to minimize doses is advocated, or chronic use of MMF or
AZA in combination with a calcineurin inhibitor is required.
T-Cell Depleting Agents
In the past, induction therapy with antilymphocyte
agents such as antilymphocyte globulin or antithymocyte
globulin or monoclonal antibody preparations such as

OKT3 was utilized immediately after liver transplantation
to rapidly induce an immune suppressed state via the rapid
destruction of the hosts T cells. However, due to significant
systemic side effects including fevers, allergic reactions,
serum sickness, and thrombocytopenia, the use of these
agents is now usually reserved for the treatment of glucocorticoid resistant rejection, or less commonly, in patients
with severe renal insufficiency in an attempt to delay the
use of either CYA or TAC, which may be associated with
worsening of renal function (41).
IL-2 Receptor Blockers
T-cell activation and proliferation following presentation of
a foreign antigen requires the induction of several cytokines, including IL-2 (interleukin 2). Antibodies directed
against the interleukin (IL)-2 receptor are effective for
initial immunosuppression, as IL-2 receptor blockade
down regulates IL-2 mediated T-cell proliferation. The IL-2

receptor antibodies such as Basiliximab and Daclizumab,
given intravenously at the time of transplant and during
the first posttransplantation week can reduce the incidence
of acute liver graft rejection when utilized in combination
with a calcineurin inhibitor, although these agents may not
be sufficient to prevent rejection when utilized alone. The
IL-2 receptor antibodies are generally well tolerated,
although side effects may include infections, gastrointestinal distress, and rarely, pulmonary edema and bronchospasm. As these agents rarely induce renal dysfunction,
many transplant programs utilize IL-2 receptor antibodies
as induction therapy in individuals with renal insufficiency at the time of transplantation (42), in an attempt to
delay initiation or diminish dose of calcineurin inhibitors,
which may exacerbate renal insufficiency.
Calcineurin Inhibitors
IL-2 inhibition effectively suppresses T-Cell activation.
Cyclosporine and TAC achieve this by binding to cytoplasmic receptors, forming complexes which inactivate calcineurin, a key enzyme in T-cell signaling. The major side
effects of both CYA and TAC include hypertension, renal
insufficiency, and neurologic complications. However,
there is evidence to suggest that obesity, hyperlipidemia,
hirsutism, and gingival hyperplasia occur more commonly
in patients who receive CYA, while a higher rate of diarrhea, insulin resistance, and diabetes is seen in patients
who receive TAC. In response to inconsistent absorption of
standard Cyclosporine, the development of a microemulsified formulation of cyclosporine (e.g., Neoral) has allowed
consistent blood levels (43). Given their efficacy and oral
administration, calcineurin inhibitors have a central role
in posttransplant immunosuppression.
Safety and efficacy of calcineurin inhibitors is generally
assessed by monitoring blood levels drawn prior to the dose
(trough), although several investigators describe that blood
levels drawn 2 h after a dose of Cyclosporine (i.e., C2 levels)
rather than trough levels more accurately indicate exposure
to drug. At many transplantation centers, the definition of
appropriate target level of calcineurin inhibitor is linked to
the patients time posttransplantation; in general, higher
levels are required in the first several months postoperatively
while the threat of rejection is acute. The target levels for
calcineurin inhibitors can be appropriate adjusted downward
as patients achieve both normal liver function and freedom
from rejection months to years following surgery. In addition, a philosophy of minimizing exposure to high levels of
calcineurin inhibitors in HBV or HCV infected patients is
adopted by many transplant centers, due to the negative
impact of overimmunosuppression on viral replication
and disease recurrence.
Antiproliferative Agents
Antiproliferative agents such as AZA and MMF prevent
the expansion of activated T cells and B cells and regulate
immune mediated injury. Azathioprine, a purine analogue,
is metabolized in the liver to its active compound, 6-mercaptopurine, which inhibits adenosine and guanine production, thus inhibiting DNA and RNA synthesis in rapidly
275
proliferating T cells. Mycophenolate Mofetil is a potent

noncompetitive inhibitor of inosine monophosphate dehydrogenase (IMPDH), an enzyme necessary for the synthesis of guanine, a purine nucleotide. Mycophenolate
Mofetil, when used in combination with a calcineurin
inhibitor and steroids has been shown to be associated
with lower rejection rates in the first 6 months posttransplantation when compared to AZA (44). The major toxicities associated with the use of either MMF or AZA are
bone marrow suppression with resultant leukopenia, anemia, and thrombocytopenia, though this is more marked
with AZA. Mycophenolate Mofetil has been associated with
a greater incidence of dyspepsia, peptic ulcers, and diarrhea when compared to AZA, while pancreatitis may occur
in individuals prescribed AZA. These side effects usually
abate by dose reduction or discontinuation. The majority of
transplant centers utilize a combination of a Calcineurin
inhibitor with either MMF or, less commonly, AZA for at
least the first 36 months posttransplantation. Since AZA
and MMF do not cause renal insufficiency, they can be
utilized in a strategy to minimize or avoid calcineurin inhibitor use, particularly in patients with renal dysfunction.
Other Immunosuppressive Agents
The limitations and untoward effects of available immunosuppressive agents have induced research and development
of alternative agents. Sirolimus (Rapamycin) (RAPA) and its
derivative Everolimus represent a new class of compounds,
which achieve their immuosuppressive effect by inhibiting
mTOR (target of Rapamycin). Inhibition of mTOR diminishes intracellular signaling distal to the IL-2 receptor and
prevents T-cell replication. As the lymphoproliferative pathways inhibited by RAPA and Everolimus are distinct from
those affected by calcineurin inhibitors, investigators have
utilized these agents in combination with calcineurin inhibitors to achieve synergistic effect. However, enthusiasm for
RAPA has been tempered by recent data showing higher
rates of hepatic arterial thrombosis in patients who receive
RAPA in the weeks immediately following transplantation
(45). In addition, impaired wound healing has been noted in
patients who receive RAPA, potentially due to impairment of
granulation mediated by inhibition of TGF-b. Leukopenia,
thrombocytopenia, and hyperlipidemia are the principal toxicities associated with RAPA and Everolimus. Recent reports
of pneumonitis in RAPA treated patients have also emerged.
A positive attribute of both RAPA and Everolimus is the
absence of renal toxicity; some data suggest that post transplantation renal insufficiency can be reversed when calcineurin inhibitors are withdrawn and RAPA is initiated (46).
Newer immunosuppressive agents will continue to be developed; it is hoped that these agents will be associated with
diminished short- and long-term toxicity and facilitate a state
of immune tolerance of the graft that will ultimately allow
minimization of the requirement for immunosuppressive
medications.
SUMMARY
Liver transplantation is the treatment of choice for appropriately selected patients with end stage liver disease.
276
Over the last several decades, significant advances in

surgical technique and immunosuppression, selection of
appropriate donors, grafts, and recipients, and improved
therapies to prevent and treat postoperative complications
have greatly improved posttransplantation outcomes.
Despite these impressive achievements, many challenges
remain. It is becoming increasingly apparent that the
growing disparity between the number of liver transplant
candidates and available organs will be associated with
escalating death rates on the transplant waiting list.
Enhanced posttransplantation survival has led to the
emergence of complications associated with patient longevity, including nonhepatic disease, complications of
immunosuppression, infections, neoplasia, and recurrence
of the primary disease for which the liver transplantation
was indicated. Further progress in liver transplantation
will be achieved by maximizing the use of available organs,
refinement and exploration of alternatives to deceased
donor liver transplantation, improvements in immunosuppression, and enhanced recognition and treatment of longterm complications, particularly recurrent liver disease.
BIBLIOGRAPHY
1. Welch CS. A note on transplantation of the whole liver in dogs.
Transplant Bull 1955;2:54.
2. Kukral JC, Littlejohn MH, Williams RK, Pancer RJ, Butz GW
Jr, Starzl TE. Hepatic function after canine liver transplantation. Arch Surg 1962;85:157165.
3. Starzl TE, Marchioro TL, Porter KA, Brettschneider L.
Related Articles, Homotransplantation of the liver. Transplantation 1967;5(Suppl):790803.
4. Mazzaferro V, Regalia E, Doci R, Andreola S, Pulvirenti A,
Bozzetti F, Montalto F, Ammatuna M, Morabito A, Gennari L.
Liver transplantation for the treatment of small hepatocellular
carcinomas in patients with cirrhosis. N Engl J Med 1996;334:
693699.
5. Moreno S, Fortun J, Quereda C, Moreno A, Perez-Elias MJ,
Martin-Davila P, de Vicente E, Barcena R, Quijano Y, Garcia
M, Nuno J. Martinez Liver transplantation in HIV-infected
recipients. Liver Transpl 2004;11:7681.
6. Kamath PS, Wiesner RH, Malinchoc M, Kremers W, Therneau
TM, Kosberg CL, DAmico G, Dickson ER, Kim WR. A model to
predict survival in patients with endstage liver disease. Hepatology 2001;33:464470.
7. Freeman RB, Wiesner RH, Edwards E, Harper A, Merion R,
Wolfe R. United Network for Organ Sharing Organ Procurement and Transplantation Network Liver and Transplantation Committee. Results of the first year of the new liver
allocation plan. Liver Transpl 2004;10:715.
8. Onaca NN, Levy MF, Sanchez EQ, Chinnakotla S, Fasola CG,
Thomas MJ, Weinstein JS, Murray NG, Goldstein RM,
Klintmalm GB. A correlation between the pretransplantation
MELD score and mortality in the first two years after liver
transplantation. Liver Transpl 2003;9:117123.
9. Merion RM, Schaubel DE, Dykstra DM, Freeman RB, Port FK,
Wolfe RA. The survival benefit of liver transplantation. Am J
Transpl 2005;5:307313.
10. http://www.UNOS.org.
11. Broelsch CE, Whitington PF, Emond JC, Heffron TG,
Thistlethwaite JR, Stevens L, Piper J, Whitington SH, Lichtor
JL. Liver transplantation in children from living related
donors. Surgical techniques and results. Ann Surg 1991;214:
428437.
12. Marcos A. Right-lobe living donor liver transplantation. Liver

Transpl 2000;6:S59S63.
13. Shiffman ML, Brown RS Jr., Olthoff KM, Everson G, Miller C,
Siegler M, Hoofnagle JH. Living donor liver transplantation:
summary of a conference at The National Institutes of Health.
Liver Transpl 2002;8:174188.
14. Emond JC, Renz JF, Ferrell LD, Rosenthal P, Lim RC, Roberts
JP, Lake JR, Ascher NL. Functional analysis of grafts from
living donors. Implications for the treatment of older recipients. Ann Surg 1996;224:544552.
15. Otero A, Gomez-Gutierrez M, Suarez F, Arnal F, FernandezGarcia A, Aguirrezabalaga J, Garcia-Buitron J, Alvarez J,
Manez R. Liver transplantation from maastricht category 2
non-heart-beating donors: A source to increase the donor pool?
Transpl Proc 2004;36:747750.
16. Ericzon BG, Larsson M, Herlenius G, Wilczek HE. Familial
Amyloidotic Polyneuropathy World Transplant Registry.
Report from the Familial Amyloidotic Polyneuropathy World
Transplant Registry (FAPWTR) and the Domino Liver Transplant Registry (DLTR). Amyloid 2003;10:6776.
17. Schemmer P, Mehrabi A, Kraus T, Sauer P, Gutt C, Uhl W,
Buchler MW. New aspects on reperfusion injury to liver
impact of organ harvest. Nephrol Dial Transpl 2004;19:2635.
18. Bhattacharjya S, Gunson BK, Mirza DF, Mayer DA, Buckels
JA, McMaster P, Neuberger JM. Delayed hepatic artery
thrombosis in adult orthotopic liver transplantation-a 12-year
experience. Transplantation 2001;71:15921596.
19. Vignali C, Cioni R, Petruzzi P, Cicorelli A, Bargellini I, Perri
M, Urbani L, Filipponi F, Bartolozzi C. Role of interventional
radiology in the management of vascular complications after
liver transplantation. Transpl Proc 2004;36:552554.
20. Lefkowitch JH. Diagnostic issues in liver transplantation
pathology. Clin Liver Dis 2002;6:555570.
21. Fondevila C, Ghobrial RM, Fuster J, Bombuy E, GarciaValdecasas JC, Busuttil RW. Biliary complications after adult
living donor liver transplantation. Transpl Proc 2003;35:
19021903.
22. Denys A, Chevallier P, Doenz F, Qanadli SD, Sommacale D,
Gillet M, Schnyder P, Bessoud B. Interventional radiology in
the management of complications after liver transplantation.
Eur Radiol 2004;14:431439.
23. Selzner N, Rudiger H, Graf R, Clavien PA. Protective strategies against ischemic injury of the liver. Gastroenterology
2003;125:917936.
24. Clavien PA, Yadav S, Sindram D, Bentley RC. Protective
effects of ischemic preconditioning for liver resection performed under inflow occlusion in humans. Ann Surg 2000;232:
155162.
25. Demetris A, Adams D, Bellamy C, Blakolmer K, Clouston A,
Dhillon AP, Fung J, Gouw A, Gustafsson B, Haga H, Harison
D, Hart J, Hubscher S, Jaffe R, Khettry U, Lassman C, Lewin
K, Martinez O, Nakazawa Y, Neil D, Pappo O, Parizhskaya M,
Randhawa P, Rasoul-Rockenschaub S, Reinholt F, Reynes M,
Robert M, Tsamandas A, Wanless I, Wiesner R, Wernerson A,
Wrba F, Wyatt J, Yamabe H. Update of the international banff
schema for liver allograft rejection: Working recommendations for the histopathologic staging and reporting of chronic
rejection. An International Panel. Hepatology 2000 Mar;
31(3):792799.
26. Neuberger J. Recurrent primary biliary cirrhosis. Baillieres
Best Pract. Res Clin Gastroenterol 2000;14:669680.
27. Wiesner RH. Liver transplantation for primary sclerosing
cholangitis: timing, outcome, impact of inflammatory bowel
disease and recurrence of disease. Best Pract Res Clin Gastroenterol 2001;15:667680.
28. Molmenti EP, Netto GJ, Murray NG, Smith DM, Molmenti H,
Crippin JS, Hoover TC, Jung G, Marubashi S, Sanchez EQ,
LUNG SOUNDS
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
Gogel B, Levy MF, Goldstein RM, Fasola CG, Gonwa TA,

Klintmalm GB. Incidence and recurrence of autoimmune/
alloimmune hepatitis in liver transplant recipients. Liver
Transpl 2002;8:519526.
Burke A, Lucey MR. Non-alcoholic fatty liver disease, nonalcoholic steatohepatitis and orthotopic liver transplantation.
Am J Transplant 2004;4:686693.
Mackie J, Groves K, Hoyle A, Garcia C, Garcia R, Gunson B,
Neuberger J. Orthotopic liver transplantation for alcoholic
liver disease: A retrospective analysis of survival, recidivism,
and risk factors predisposing to recidivism. Liver Transpl 2001;
7:418427.
Berenguer M, Prieto M, Rayon J, Mora J, Pastor M, Vicente O,
et al. Natural history of clinically compensated hepatitis C virus
related graft cirrhosis after liver transplantation. Hepatology
2000;32:852858.
Gane E. The natural history and outcome of liver transplantation in hepatitis C virusinfected recipients. Liver Transpl
2003;9:S28S34.
Garcia-Retortillo M, Forns X, Llovet JM, Navasa M, Feliu A,
Massaguer A, Bruguera M, Fuster J, Garcia-Valdecasas JC,
Rimola A. Hepatitis C recurrence is more severe after living
donor compared to cadaveric liver transplantation. Hepatology 2004;40:699707.
Gaglio PJ, Malireddy S, Levitt BS, Lapointe-Rudow D, Lefkowitch J, Kinkhabwala M, Russo MW, Emond JC, Brown RS Jr.
Increased risk of cholestatic hepatitis C in recipients of grafts
from living versus cadaveric liver donors. Liver Transpl
2003;9:10281035.
Shiffman ML, Stravitz RT, Contos MJ, Mills AS, Sterling RK,
Luketic VA, Sanyal AJ, Cotterell A, Maluf D, Posner MP,
Fisher RA. Histologic recurrence of chronic hepatitis C virus in
patients after living donor and deceased donor liver transplantation. Liver Transpl 2004;10:12481255.
Russo MW, Galanko J, Beavers K, Fried MW, Shrestha R.
Patient and graft survival in hepatitis C recipients after adult
living donor liver transplantation in the United States. Liver
Transpl 2004;10:340346.
Crippin JS, McCashland T, Terrault N, Sheiner P, Charlton
MR. A pilot study of the tolerability and efficacy of antiviral
therapy in hepatitis C virus-infected patients awaiting liver
transplantation. Liver Transpl 2002;8:350355.
Everson GT. Treatment of chronic hepatitis C in patients with
decompensated cirrhosis. Rev Gastroenterol Disord 2004;4:
S31S38.
Gane E. Treatment of recurrent hepatitis C. Liver Transpl
2002;8:S28S37.
Conti F, Morelon E, Calmus Y. Immuosuppressive therapy in
liver transplantation. J Hepatol 2003;39:664678.
Tector AJ, Fridell JA, Mangus RS, Shah A, Milgrom M, Kwo P,
Chalasani N, Yoo H, Rouch D, Liangpunsakul S, Herring S,
Lumeng L. Promising early results with immunosuppression
using rabbit anti-thymocyte globulin and steroids with
delayed introduction of tacrolimus in adult liver transplant
recipients. Liver Transpl 2004;10:404407.
Liu CL, Fan ST, Lo CM, Chan SC, Ng IO, Lai CL, Wong J.
Interleukin 2 receptor antibody (basiliximab) for immunosuppressive induction therapy after liver transplantation: A protocol with early elimination of steroids and reduction of
Tacrolimus dosage. Liver Transpl 2004;10:728733.
Lilly LB, Grant D. Optimization of cyclosporine for liver
transplantation. Transpl Proc 2004;36:267S270S.
Wiesner R, Rabkin J, Klintmalm G, McDiarmid S, Langnas A,
Punch J, McMaster P, Kalayoglu M, Levy G, Freeman R,
Bismuth H, Neuhaus P, Mamelok R, Wang W. A randomized
double-blind comparative study of mycophenolate mofetil
and azathioprine in combination with cyclosporine and
277
corticosteroids in primary liver transplant recipients. Liver

Transpl 2001;7:442450.
45. Trotter JF. Sirolimus in liver transplantation. Transplant
Proc 2003;35:193200.
46. Nair S, Eason J, Loss G. Sirolimus monotherapy in nephrotoxicity due to calcineurin inhibitors in liver transplant recipients. Liver Transpl 2003;9:126129.
See also DIFFERENTIAL
COUNTS, AUTOMATED; PHARMACOKINETICS AND
PHARMACODYNAMICS; IMMUNOTHERAPY.
LONG BONE FRACTURE. See BONE UNUNITED

FRACTURE AND SPINAL FUSION, ELECTRICAL TREATMENT OF.
LUNG MECHANICS. See RESPIRATORY MECHANICS AND

GAS EXCHANGE.
LUNG PHYSIOLOGY.
See PULMONARY
PHYSIOLOGY.
LUNG SOUNDS
ROBERT G. LOUDON
RAYMOND L. H. MURPHY
INTRODUCTION
Medical devices and instrumentation have developed rapidly
in the last few decades, yet the first diagnostic medical
instrument, the stethoscope, is still the most widely used,
and it has changed only superficially in design and function.
The lungs, as we breathe, produce sounds that are
transmitted to the body surface and to the mouth. The
characteristics of these sounds convey information about
the sound-producing and -transmitting structures. This
information often has diagnostic value. Auscultation of the
lungs is therefore widely taught and practiced. Textbooks of
physical diagnosis present a body of information that has
been derived by careful workers since the introduction of
the stethoscope by R.T.H. Laennec in 1819. Much of that
information was indeed presented by Laennec himself in
his remarkable treatise, De lAuscultation Mediate(1,2).
In this article, the medical devices and instruments that
have been applied to the study of lung sounds, including
the traditional acoustic stethoscope are reviewed. This
survey will include sound transducers and their placement, methods, and equipment used for the recording
and analysis of lung sounds, results obtained by the use
of these techniques, and their clinical meaning. Recent
work on this subject helps in the understanding of what
we hear with the stethoscope; some is aimed at answering
specific questions in physiology or pathology, and some is
designed to provide new diagnostic and monitoring tools.
Much of this work has been done in the past three decades,
reflecting the enormous increase in the availability and
quality of sound recording and processing techniques during that period. Reviews of lung sounds (35) and the
success of the International Lung Sounds Association
and its annual meetings bear witness to the upsurge of
interest in the subject. Recommended standards for terms
and techniques used in computerized respiratory sound
278
LUNG SOUNDS
analysis (CORSA) have been prepared by a Task Force of

the European Respiratory Society and published in the
European Respiratory Review series (6). Better understanding of the meaning of current and future observations
promises a larger place in the future for clinical and
research applications.
THE STETHOSCOPE
The introduction of the stethoscope is an interesting story,
well described in a bicentenary appreciation of Laennecs
birth (7). Laennec, a young physician practicing in Paris,
had occasionally found it useful to listen directly to a
patients chest, as had been done by physicians at least
since the time of Hippocrates. In 1816, he wished to listen to
an obese young ladys heart, but was reluctant to do so. He
recollected (and this part of the story may be apocryphal)
having seen boys playing on a park bench, one listening to the
wooden bench at one end with his ear, and the other scratching the other end. Laennecs own words were that he
happened to recollect a simple and well know fact in acoustics, that sound could be transmitted through solid material
or along a tube. He rolled a quire of paper into a sort of
cylinder, placed one end over her heart, and listened at the
other end. He was not a little surprised and pleased to hear
the sounds more clearly in this mediate fashion than he had
ever been able to do by the immediate application of his ear
(2). Over the next 3 years he amassed an enormous amount of
information about the sounds heard over the chests of his
patients. As he did all of the autopsies at the Hopital Necker
in Paris where he worked, he could often relate these sounds
to the underlying pathology.
The first edition of Laennecs book (1) cost 13 francs for
the two volumes; for an extra 2.50 francs, one received a
wooden stethoscope. This cylinder served its purpose
well. Modifications were introduced over the years, such
as earpieces, flexible tubing, binaural stethoscopes, and a
diaphragm on the chest piece. The relative merits of diaphragm and bell, the effect of the length and bore of the
tubing, and the convenience of different patterns have been
debated over the years, and the design of modern stethoscopes has been largely empirical, better models surviving
because of their popularity with auscultators. Some characteristics that acousticians might think of as defects may
indeed be advantageous from the physicians point of view.
Those using them tend to feel comfortable listening to
sounds with which they are familiar and may reject a
stethoscope that lets them hear too much.
The assessment of acoustical performance of stethoscopes is not as simple as it might seem, and approaches
to this problem have been described by several authors
(810). The value of the traditional stethoscope is in no way
reduced by the recent introduction of devices and instruments that can record and analyze the sounds that we hear.
Rather, its value is increased. Appropriate use on new
medical devices and instrumentation adds science to art,
measurement to impression, and recordings to memory.
Better understanding of what lung sounds mean, and of
how much the simple stethoscope can tell us and how much
it cannot, will make the use of the simple stethoscope in
examining rooms or on clinical rounds more important

than ever.
SOUND TRANSDUCERS
Microphones transform mechanical energy to electrical
energy, in the sound frequency range. Mechanical movement at the chest wall, resulting from the transmission of
vibrations representing lung sounds to the chest wall surface, may be detected by any one of several devices. The
main categories are ceramic, condenser (capacitor), and
electret microphones. Ceramic microphones use a piezoelectric ceramic element that produces voltage when it is
stressed. They tend to be stable and rugged and do not need
a bias voltage for operation. Condenser microphones of the
conventional type act as a variable capacitor that requires
a bias voltage. They have good sensitivity and frequencyresponse characteristics. Electret microphones are a more
recent type; a permanent charge on the diaphragm and no
free electrostatic charge on its surface relieve the need
for a polarizing (bias) voltage and reduce sensitivity to
humidity.
Most microphones are designed to receive sound transmitted through air. Air coupling has been used by several
investigators recording sounds from the surface of the
chest wall, or from the trachea, and it is not surprising
that stethoscope chest-pieces have been used for this purpose. The sound transmitted through the air column in
stethoscope tubing can be applied to a microphone just as it
can to an auscultating eardrum. Direct mechanical coupling of the transducer to the signal site (chest wall or
tracheal surface) is an alternative to air coupling.
Several authors have reviewed the relative advantages
and disadvantages of the various types of microphones as
lung sound transducers (11,12). Desirable characteristics
include sensitivity, rejection of ambient noise and surface
noise, appropriate frequency response, insensitivity to variation in pressure of application, ease of attachment, ruggedness, and low price. Sensitivity is necessary because of
the low level of the sound signal. Vesicular breath sounds
will on occasion be virtually inaudible, for example, when
airflow rates at the mouth are <0.27 L/s (13).
It is not always possible to study lung sounds in ideal
circumstances, and rejection of ambient noise is important
for many applications. Microphone housing can be helpful
in this regard. Heart sounds are often of greater amplitude
than lung sounds and may obscure them. They can be made
less troublesome by the frequency response of the microphone because heart sounds are in a lower frequency
range. Air coupling or inherent microphone characteristics
may help by increasing the high frequency response.
Microphone placement can also reduce the interference
from heart sounds, which are, of course, loudest over the
front of the chest, particularly in the left lower zone, and
are less obtrusive on the right side, especially at the base of
the right lung posteriorly. One method that has been
adopted to reduce contamination of lung sounds by the
heart sounds is to record the electrocardiogram simultaneously and to use some form of gating to delete segments
where the heart sounds are present (14). The periodicity of
LUNG SOUNDS
the heart sound makes this an attractive alternative for

some purposes. Muscle noise can also contaminate lung
sounds; again the frequency content of muscle noise is
considerably lower than that of lung sounds, and a microphone that is insensitive to low frequency noise, or subsequent filtration of the signal, can be helpful. Muscle noise
has the disadvantage of being timed with respiration
because it arises from respiratory muscle activity, and this
prevents it from being gated out on a time base. Most
investigators have found that the frequency range of most
interest in the recording and analysis of lung sounds lies
between 100 and 1000 Hz, well within the frequency range
of most microphones.
Surface noise is another important source of difficulty
that can arise in recording and interpreting lung sounds.
The movements associated with respiration make it easy
for the microphone to slide over the skin surface in phase
with respiration, producing sounds that are in phase with
respiration, may be in the same frequency range as lung
sounds, and may be very difficult to distinguish from
friction sounds such as a pleural friction rub. Surface noise
is more likely to arise when the microphone is mechanically
in contact with, but not firmly fixed to, the chest wall. Air
coupling may have advantages over mechanical coupling in
this respect, but not always if the chest piece is of the
diaphragm type commonly used in stethoscopes. Respiratory movement may also cause changes in the pressure
with which a microphone is applied to the chest wall; if the
microphone is strapped to the chest by a circumferential
band, pressure on the microphone will increase as inspiration occurs and the chest diameter increases. Variation in
pressure of the microphone against the chest wall is liable
to alter the acoustic coupling, particularly if mechanical
coupling is used to transmit surface movement to the
sensitive microphone element. If the pressure exerted is
sufficient, the deformation of the sensitive element may
approach the limit of its range, damping the signal. Aircoupled microphones are less sensitive to changes in pressure of application, provided that the air chamber between
chest wall surface and microphone element is vented to
theoutside, usually by a small-bore needle; but too large a
vent may increase the amount of ambient sound recorded
(15).
For some purposes, the sound transducer is applied only
briefly at a specific site on the chest wall while a few
breaths are recorded. For monitoring purposes, attachment of a sound recording device for a period of hours or
overnight may be necessary. Lightness and small bulk are
important in this type of application, and in some cases
two-sided adhesive tape or an adhesive patch similar to
that used for electrocardiograph electrodes is adequate for
attachment.
If chest wall surface movement is unimpeded, the vibrations that correspond to the lung sound do not involve
actual mechanical displacement of the chest wall surface
by more than a few micrometers. A sensor applied to the
surface may measure displacement or, if it applies a load to
the chest wall surface, it may measure pressure rather
than displacement, or a combination of the two. Some
sound transducers measure acceleration rather than
actual physical displacement. In each case, the reaction
279
of the sensor to the signal being sensed will influence its

characteristics. Inertia, rigidity, or counterpressure by the
sensing element may cause distortion of the sound. Particularly in the case of accelerometers, the mass of the
sensing element will determine its frequency response
characteristics. It is not always clear what criteria are
used in making a decision about microphone type. The
human ear is remarkably good at separating out the different sounds that may be combined to form a mixed signal,
and often the final judgment may be made by listening to
replay of a recorded signal. The efficiency of a particular
sound system depends on the purpose for which it is
intended, but unless the signal is listened to with an
educated ear it is easy to be misled by, for example,
frequency components whose origin is not obvious from
inspection of a graphic or calculated spectrum.
RECORDING AND DISPLAY SYSTEMS

Those using devices and instruments to study lung sounds
will choose recording and display systems appropriate to
their purpose. Audio tape and strip-chart recorders have
now virtually all been replaced by computers or systems
designed or modified for the purpose. The signals of interest may be presented to the observer audibly, visually, or in
a variety of forms during and after analysis. Standard
physical examination of the chest does, in a sense, present
audible and visual displays to the clinician. The stethoscope presents an audible signal at the earpieces, and the
clinician observes his patient breathe to get a visual display
of respiratory movement.
For teaching purposes at the bedside, an electronic stethoscope or microphone may be connected to several headsets
worn by students, by telemetry if preferred, giving the instructor an opportunity to share the sounds with them. In this way,
a realistic learning experience is provided with less imposition
on the patients patience. Recording of sounds for teaching
purposes usually involves a computer system, or an electronic
stethoscope and audio tape recorder. Standard audiovisual
equipment has been used for editing, for adding comments,
and for preparation of cassettes or disks for distribution (16,17)
for teaching purposes.
For research purposes, arrays of microphones are now
available with computer recording, analysis, and display
systems to show the distribution of sound signals over the
surface of the chest (1820). Brief differences in time of
sound signals have clinical relevance by allowing comparison in timing of the same sound signal of a crackle or the
start of a wheeze arriving at different surface sites in the
same patient. And on a longer time base, in asthmatics, for
example, the site, the frequency pattern, and the sound
amplitude of wheezing may change during exercise, sleep,
exposure to cold air or to inhalants such as pollen, or
industrial exposure, or in response to drug treatment.
Sleep disorders such as nocturnal asthma, the sleep apnea
syndrome, and snoring, may be studied by sound monitoring. Nocturnal asthma and snoring are present in the same
patient more often than would be expected as a result of
chance alone, especially in asthmatics under the age of 40
(21). Snoring is a respiratory, but not a lung sound, as it
280
LUNG SOUNDS
rises in the upper airways, at or above the larynx. Possible

explanations for the association with asthma, and sound
monitoring methods and devices, have been reviewed (22).
Comparisons over long periods of time were once made
by recording the results of analyses of wheezes, rather than
by comparing the actual recorded sounds. The development
and proliferation of computers with rapidly increasing
audiovisual capability and storage capacity are now, however, changing the situation to allow storage of original
data on tape or disk together with derived values. Kraman
et al. (23) evaluated minidisk recorders, with their considerable increase of storage capacity for music, for lung
sound recording. They found no distortion of frequency or
waveforms that would interfere with this use. For some
studies, analyzing sound signals in real time as they are
being acquired makes it simpler to monitor results as they
accrue, and helps direct the course of an experiment.
An early example of audiovisual recording is in a paper by
Krumpe et al. (24), in which the authors discuss the evaluation
of bronchial air leaks by auscultation and phonopneumography. They describe three patients who develop air leaks from
the bronchi after resectional lung surgery and in whom
videophonopneumography provided more precise correlation
of abnormal sounds with the underlying visibly leaking bronchial abnormalities. Audiovisual tapes or disks are useful for
teaching or demonstration purposes, by providing examples of
classical or of unusual sounds.
Simultaneous sound recordings at several sites have
been used to study the spatial distribution of lung sounds.
This has provided information on regional ventilation, and
on the localization of abnormalities in disease such as
pneumonia, airways obstruction, bullae, or small areas
of infarction, atelectasis, fibrosis, or interstitial lung disease. Indeed, lung imaging by sound production provides a
potential alternative to chest X rays and computed tomography (CT) scans, without the need to inject possibly
damaging energy or drugs.
For research purposes, analysis of sound signals and any
associated physiological measurements were formerly conducted off-line. The signals were recorded on tape or disk
and replayed for analysis. This allowed editing for selection of
relevant segments of data and for quality control and signal
conditioning, such as amplification, filtering, or attenuation.
The purpose of each study will determine the equipment
needs, but most current lung sound research uses computers
with high speed audiovisual capabilities. These can be adapted
to record lung sounds along with physiological respiratory
variables, such as airflow, lung volume, and esophageal pressure, measured simultaneously, which can then be related to
the lung sounds. If relationships in time are to be studied with
any precision, it is necessary to record signals together on one
medium and it is necessary to know the frequency characteristics of the items of equipment used, and the time delays
introduced by filters, envelope detectors, integrators, frequency analyzers, and other acquisition or processing devices.
SOUND ANALYSIS
Sound amplitude and frequency content are the two measurements that most commonly form the basis of lung
sound analysis systems. Early studies presented the sound

signal as a time-amplitude plot. If such plots represent a
respiratory cycle on a few centimeters of paper, the result is
a compressed representation that superimposes many successive sound signal cycles to form an envelope. Simple
integrating and rectifying circuits can provide the outline
of the envelope as a single line, thus acting as an envelope
detector, acdc converter, or sound-level meter. Filters
incorporated in such circuitry can yield a method for
comparing sound amplitude in different frequency bands
(14,17) or to provide a signal believed to represent the
important band range of vesicular sound from the ventilation point of view (25).
The sound spectrogram is really an extension of this
principle, the signal of interest being passed repetitively
through a narrow bandpass filter with slowly changing
center frequency and the signals passed being assembled to
present a graphic display of time on the horizontal axis,
sound frequency on the vertical axis, and sound amplitude
by the degree of blackening of the paper. Sound spectrograms of this type, used routinely in the speech sciences,
were applied to heart and lung sounds extensively by
McKusick et al. (26) and are still widely used to good effect.
The time-amplitude plot of a sound signal has been used
to advantage in a different way by Murphy et al. (27).
Features of the sound waveform cannot be studied in detail
without using a rapid time sweep on an oscilloscope, and
only a brief (a few milliseconds) segment can be viewed in
this way. By digitizing a sound signal at a rapid rate and
playing the signal back through a digital-to-analogue converter (DAC), a time-expanded waveform was prepared.
This has proved of particular value in studying crackles
(rales), the brief sounds heard over fibrotic, edematous,
consolidated, or atelectatic lung. Measurable characteristics of these crackles, such as the initial or the largest
deflection width, show diagnostic value and automatic
methods for their measurement are now being applied.
The sound characteristics of rhonchi, as opposed to
crackles (continuous versus discontinuous adventitious
sounds) require an additional approach. Essentially, they
are longer in duration, possessed of perceptible pitch, and
have a repetitive waveform pattern. Waveform analysis is
a rapidly moving field. Sound frequency spectrum analysis
of lung sounds has most frequently been reported in terms
of discrete Fourier analysis. Several workers have used a
fast Fourier transform algorithm to measure frequency
content of signal segments. One way of representing
time-variant sound signals is to assemble a sequence of
spectra with frequency on the horizontal axis, sound amplitude or power on the vertical axis, and time on an oblique
axis. Usually, some overlapping of the sequential segments
and appropriate windowing (e.g., Hanning) are used. The
resulting birds-eye view has proved to be readily related
to sounds, providing a mental image that can evoke a
mental image of the sounds represented. Individual peaks
on a frequency spectrum may be related to individual
wheezes coming from the chest, and peak detection programs have been used (28,29) to compare them statistically. The fast Fourier transform is the most frequently
reported type of waveform analysis, but other techniques,
such as those of linear predictive coding (LPC), the
LUNG SOUNDS
maximal entropy method of waveform analysis, fractaldimension analysis, wavelet networks, and artificial
neural networks, are being explored. They are most likely
to prove useful in brief sounds, in timing the onset or rapid
changes in complex sounds, or in noting time relationships
among sounds recorded at separate or at adjacent sensors.
Any graphic form of waveform analysis is more readily
interpreted when it can be combined with visual examination of a simultaneous time-amplitude plot.
RESULTS AND CLINICAL APPLICATIONS

Increasing attention and techniques for more exact representation have led to a rapid growth in information available about lung sounds. The meaning of these various
items of information will emerge more slowly, as will
clinical applications. The objective, quantitative study of
lung sounds, is still at an interesting rapid growth phase of
development. It is clear that a good deal of information is
contained in the signals that we hear emerging from the
chest (2) and that auscultation is one of the safest of
diagnostic procedures, since no external energy or chemical is inserted into the body. It is also clear that some of the
information conveyed would be difficult to obtain in any
other way. Much of it is regional or local and may be able to
tell us about mechanical events and structural characteristics at specific sites in the chest (30,31). The vesicular
lung sounds have been studied in sufficient detail that we
now know more about the probable general range of bronchial dimensions involved in the production of these
sounds, but not the exact site; the effects of flow rate
and lung volume, but not the exact nature of the relationships; and we know that there are relationships between
vesicular lung sound intensity and regional ventilation,
but not their exact nature. The roles of production and of
transmission of these sounds are not always easy to distinguish from one another in the end-product, sensed at the
site of their detection, but recent work by Kiyokawa and
Pasterkamp (32) shows progress in this distinction.
We know that wheezing indicates airflow obstruction
and roughly its levels in the bronchial tree. We know that
several factors, such as airway dimensions, geometry, and
compressibility, are important. Endobronchial surface
characteristics and the presence and nature of secretions
may also have some effect. We know that flow rates and
intrathoracic pressure and volume history affect wheezes;
but we do not know the relative importance of these factors
and the extent of variation from one disease state to
another. Crackles are known to be associated with certain
diseases and not with other radiographically similar diseases, but we are not sure why. We know that crackles from
different types of abnormal lungs have different characteristics, but a great deal of clinical observation will be needed
to test their diagnostic value: and physiological or pathological studies to understand the basic mechanisms
involved.
Laennecs stethoscopeand for that matter the stethoscope pulled currently from the pocket of a white coat
allows the user to consider the sound of one breath at one
place at one time. Medical devices and equipment are now
281
being developed that can expand the observations in time,

in space, in content, and in information; for example, from
one or two breaths to hundreds of breaths, and from one
specific point on the chest to the entire chest. From one
breath described or remembered as vesicular, reduced in
volume, with a few end-expiratory crackles the information
may expand to an assembly of pages of tables and graphs
showing a variety of measured features. These can include
diagrams of the chest showing where and how the lungs
and ventilation vary, where and how much airflow obstruction or lung collapse is present, and can offer a regional
description of airways diameters and other characteristics.
Que et al. (33) developed a system to measure tracheal
flow from tracheal sounds, and to use this to estimate tidal
volume, minute ventilation, respiratory frequency, mean
inspiratory flow rate, and duty cycle. Careful observations
and comparison of the results with simultaneously
recorded pneumotachygraph-derived volumes in various
postures allowed them to address the problems inherent in
the adverse signal/noise ratio and the low level of the flowderived sound at flow rates seen in quiet breathing. The
system that they developed suggests that their method of
phonospirometry measures overall ventilation reasonably
accurately without mouthpiece, noseclip, or rigid postural
constraints.
The study by Kiyokawa and Pasterkamp (32) in a sense
complements this by measuring lung sounds at two closely
spaced sensors on the chest surface. In five healthy subjects, volume-dependent variations in phase and amplitude
of signals recorded over the lower lobe might reflect spatial
variations of airways and diaphragm during breathing.
These authors noted similar variations in phase and amplitude on passive sound transmission, suggesting that a difference in sound transmission was a more likely cause of the
variations than a difference in sound generation. Their observations compare local sounds that reflect local circumstances;
the observations discussed in the previous paragraph concern
central sounds that reflect total ventilation.
Several systems are now available or under development that can record sound signals simultaneously from a
number of sites, with or without associated physiological
signals, and present the observations for read-out by the
physician. Lung sound documentation and analysis can
now be done on personal digital assistants (PDAs) as well
as on laptop computers. Stethoscopes can be connected to
these devices wirelessly or by a short cable. This allows
objective quantification of these sounds at the bedside
(34,35). A personal computer based telemedicine system
has been described in which two remote hemodialysis sites
were connected by high speed telephone lines to allow video
and audio supervision of dialysis from a central site (36).
Such equipment may eventually be used to supplement
or perhaps in some circumstances replaceother diagnostic devices such as fluoroscopy or other types of radiographic imaging. They have the great advantage of
avoiding the subjection of a patient to any potentially
harmful radiation or other energy, and can therefore be
used over prolonged periods of time.
Transthoracic speed of sound introduced at the mouth
or the supraclavicular space (35) can be mapped at several sites on the chest using sound input with specific
282
LUNG SOUNDS
characteristics. This may allow noninvasive monitoring of

conditions such as pneumonia, congestive heart failure,
or pleural effusion that increase intrathoracic density.
Chronic obstructive lung disease may be detected by
reading lung sound maps showing time intensity plots
at several sites over the chest; this appears to be more
accurate than current clinical diagnostic methods. The
ability to detect diaphragmatic movement by multichannel lung sound analysis suggests that it may prove to be
an inexpensive bedside test. It may also have useful
applications in ventilator management.
It seems clear that wider application of these new
developments in lung sound analysis will lead to safe,
useful, and rewarding forms of clinical and physiological
information that can answer many imaging, diagnostic,
and monitoring problems.
BIBLIOGRAPHY
1. Laennec RTH. De lauscultation mediate ou traite du diagnostic des maladies des poumon et du coeur, fonde principalement sur ce nouveau moyen dexploration. 1st French ed.,
Volumes 2, Paris: Brosson et Chaude; 1819.
2. Laennec RTH, trans. by Forbes J 1st American ed. Philadelphia: James Webster; 1823. p 211.
3. Mikami R, Murao M, Cugell DW, Chretien J, Cole P, MeierSydow J, Murphy RL, Loudon RG. International symposium on
lung sounds. Synopsis of proceedings. Chest 1987;92:342345.
4. Bettencourt PE, Del Bono EA, Spiegelman D, Herzmark E,
Murphy RL. Clinical utility of chest auscultation in common
pulmonary diseases. Am J Respir Crit Care Med 1994;150:
12911297.
5. Pasterkamp H, Kraman SS, Wodicka GR. Respiratory sounds.
Advances beyond the stethoscope. Am J Respir Crit Care Med
1997;156:974987.
6. Sovijarvi AHA, Vanderschoot J, Earis JE. Computerized
Respiratory Sound Analysis (CORSA): Recommended standards for terms and techniques. Eur Respir Rev 2000;10:
77:585649.
7. Sakula A. Laennec RTH 17811926. His life and work. A
bicentenary appreciation. Thorax 1981;36:81.
8. Ertel PY. Stethoscope acoustics and the engineer: Concepts
and problems. J Audio Eng Soc 1971;19:182188.
9. Charbonneau G, Sudraud M. Measurement of the frequency
response of some commonly used stethoscopes. Consequences
to cardiac and pulmonary auscultation. Bull Eur Physiol
1985;21:4955.
10. Abella M, Formolo J, Penney DG. Comparison of the acoustic
properties of six popular stethoscopes. J Acoust Soc Am
1992;91:22242228.
11. Charbonneau G, Racineux JL, Sudraud M, Tuchais E. An
accurate recording system and its use in breath sounds spectral analysis. J Appl Physiol 1983;55:11201127.
12. Pasterkamp H, Kraman SS, DeFrain PD, Wodika GR. Measurement of respiratory acoustical signals. Comparison of
sensors. Chest 1993;104:15181993.
13. Kraman SS. Lung sounds: Relative sites of origin and comparative amplitude in normal subjects. Lung 1983;161:5764.
14. Pasterkamp H, Fenton R, Tal A, Chernick V. Interference of
cardiovascular sounds with phonopneumography in children.
Am Rev Respir Dis 1985;131:6164.
15. Kraman SS, Wodicka GR, Oh Y, Pasterkamp H. Measurement
of respiratory acoustic signals. Effect of microphone air cavity
width, shape, and venting. Chest 1995;108:10041008.
16. Cugell DW. Use of tape recordings of respiratory sound and

breathing pattern for instruction in pulmonary auscultation.
17. Banaszak EF, Kory RC, Snider GL. Phonopneumography. Am
Rev Respir Dis 1973;107:449455.
18. Kompis M, Pasterkamp H, Wodicka GR. Acoustic imaging of
the human chest. Chest 2001;120:13091321.
19. Sun X, Cheetam BM, Earis JE. Real time analysis of lung
sounds. Technol Health Care 1998;6:1122.
20. Bergstresser T, Ofengeim D, Vyshedskiy A, Shane J, Murphy
R. Sound transmission in the lung as a function of lung
volume. J Appl Physiol 2002;93:667674.
21. Fitzpatrick MF, Martin K, Fossey E, Shapiro CM, Elton RA,
Douglas NJ. Snoring, asthma and sleep disturbance in Britain: a community-based survey. Eur Respir J 1993;6:531535.
22. Dalmasso F, Prota R. Snoring: analysis, measurement, clinical
implications and applications. Eur Respir J 1996;9:146159.
23. Kraman SS, Wodicka GR, Kiyokawa H, Pasterkamp H. Are
minidisc recorders adequate for the study of respiratory
sounds?. Biomed Instrum Technol 2002;36:177182.
24. Krumpe PE, Hadley J, Marcum RA. Evaluation of bronchial
air leaks by auscultation and phonopneumography. Chest
1984;85:777781.
25. Ploysongsang Y, Martin RR, Ross RD, Loudon RG, Macklem
PT. Breath sounds and regional ventilation. Am Rev Respir
Dis 1977;116:187199.
26. McKusick VA, Jenkins JT, Webb GN. The acoustic basis of the
chest examination: Studies by means of sound spectrography.
Am Rev Tuberc 1955;72:122134.
27. Murphy RLH, Holford SK, Knowler WC. Visual lung sound
characterization by time-expanded wave-form analysis. N
Engl J Med 1977;296:968971.
28. Baughman RP, Loudon RG. Lung sound analysis for continuous evaluation of airflow obstruction in asthma. Chest
1985;88:364368.
29. Pasterkamp H, Tal H, Leahy F, Fenton R, Chernik V. The
effect of anticholinergic treatment on post exertional wheezing
in asthma studied by phonopneumography and spirometry.
30. Nath AR, Capel LH. Inspiratory crackles and the mechanical
events of breathing. Thorax 1974;29:695698.
31. Murphy RLH, Jr., Gaensler EA, Holford SK, Delbono EA,
Eppler G. Crackles in the early detection of asbestosis. Am
Rev Respir Dis 1984;129:375379.
32. Kiyokawa H, Pasterkamp H. Volume-dependent variations of
regional lung sound, amplitude, and phase. J Appl Physiol
2002;93:10301038.
33. Que C-L, Kolmaga C, Durand L-G, Kelly SM, Macklem PT. Phonospirometry for noninvasive measurement of ventilation: Methodology and preliminary results. J Appl Physiol 2002;93:15151526.
34. Bergstresser T, Ofengeim D, Vyshedskiy A, Shane J, Murphy
R. Sound transmission in the lung as a function of lung
volume. J Appl Physiol 2002;93:667674.
35. Paciej R, Vyshedskiy A, Shane J, Murphy R. Transpulmonary
speed of sound input into the supraclavicular space. J Appl
Physiol 2002;94:604611.
36. Winchester JF, Tohme WG, Schulman KA, Collman J, Johnson A, Meissner MC, Rathore S, Khanafer N, Eisenberg JM,
Mun SK. Hemodialysis patient management by telemedicine:
Design and implementation. ASAIO J 1997;43:M763766.
See also PULMONARY
PHYSIOLOGY; RESPIRATORY MECHANICS AND GAS
EXCHANGE.
LVDT. See LINEAR VARIABLE DIFFERENTIAL TRANSFORMERS.
M
MAB. See MONOCLONAL ANTIBODIES.
magnetic fields to be created using superconducting

magnets, whereas smaller fields are possible with
permanent magnets. In most conventional systems, the
static field is aligned along the longitudinal axis or the long
axis of the body, as shown in the z axis in Fig. 1. Clinical MRI
scanners have been built with static fields ranging from
0.1 to 7 T, but the vast majority of scanners are between
0.5 and 3.0 T.

W. F. BLOCK
A. L. ALEXANDER
S. B. FAIN
M. E. MEYEREND
C. J. MORAN
S. B REEDER
K. K. VIGEN
O. WIEBEN
University of
WisconsinMadison
Milwaukee
Madison, Wisconsin
THEORY
Creating Net Magnetization
Consider a static field oriented along the z axis with magnitude B0, or represented as a vector B B0k. Hydrogen
protons have a quantum operator whose z component is
quantized to . According to quantum mechanics, only
two discrete sets of orientations exist for the magnetic
dipole of each hydrogen nucleus. In the parallel energy
state, the magnetic moment vector m orients itself so that
its projection on the z axis aligns with the direction of the
main magnetic field B0. In the antiparallel energy state,
this projection aligns in the opposite direction of the
main field. It can be shown that the two allowed angles
between magnetic dipoles and the static field are u 548
(1), and thus a population of spins will be oriented as in
Fig. 2b.
The ratio of spins in the parallel state n to the spins in
antiparallel state n is given by the Boltzmann equation
J. H. BRITTAIN
General Electric Healthcare
INTRODUCTION
The principle of nuclear magnetic resonance (NMR) was
discovered by Felix Bloch and Edward Purcell independently in 1946. The two were awarded the Nobel Prize in
Physics for the discovery, which had numerous applications in studying molecular structure and diffusion. Atomic
nuclei with an odd number of protons or an odd number of
neutrons behave like spinning particles, which, in turn,
create a small nuclear spin angular momentum. This
angular momentum of an electrically charged particle such
as the nucleus of a proton leads to a magnetic dipole
moment. In the absence of an external magnetic field,
the orientation of these magnetic moments is random
due to thermal random motion. These magnetic moments
are referred to as spins, because the fundamentals of the
phenomena can be explained using classical physics where
the moments act similarly to toy tops or gyroscopes. The
NMR phenomenon exists in several atoms and is used
today to study metabolism via imaging. However, hydrogen is the simplest and most imaged nucleus in MR examinations of biological tissues because of its prevalence and
high signal compared with other nuclei.
NMR imaging was renamed Magnetic resonance
imaging (MRI) to remove the word nuclear, which the
general public associated with ionizing radiation. MRI
can be explained as the interaction of spins with three
magnetic fields: a large static field referred to as B0, which
organizes the orientation of the spins; a radio frequency
(RF) magnetic field referred to as B1, which perturbs the
spins so that a signal can be created; and spatially varying
magnetic fields referred to as gradients, which encode the
spatial location of the spins. These subsystems are shown in
Fig. 1.
When an external magnetic field is present, the distribution of the magnetic moments is no longer random.
Current technology allows large, homogenous static
g2p
DE
n
e kT e h B0 kT
(1)
where g is a nuclei-specific constant referred to as the

gyromagnetic ratio, k denotes the Boltzmann constant,
and T is the absolute temperature. There are only slightly
more spins in the parallel state than in the antiparallel
state because this state is of lower energy; however, the
prevalence of water in biological tissue can create an
adequate signal with this differential. This distribution
of spin orientations in a small volume element results in
an average or net magnetization M, which aligns along
the longitudinal or z axis, as shown in Fig. 2b. The entire
process is referred to as polarization. The contributions in
the transverse (xy) plane sum to zero. As the argument
of the exponential in Equation 1 is small and the
difference in energy levels varies proportionally with field
strength, the length of the net magnetization vector varies
linearly with field strength. A quantum mechanics
description of the spin distribution can be found
elsewhere (2).
Signal Generation
The behavior of the net magnetization vector in an external
field can be described by the classical model according to
283
284
Superconducting magnet
Y gradient coil
Figure 1. Clinical 1.5 T MRI scanner with static
field oriented along long axis of the body (z).
Patients head lies in RF coil, which is used to
both perturb and receive MR signal. Scanner
bed will move patient into middle of cylinder
before imaging begins. MR gradient coils for y
dimension are shown, which have mirrored coils
on the opposite side of the magnet. A portion of the
z gradient, based on solenoid design, is also shown.
y
z
x
RF Transmitter and Receiver
B0
the Bloch equation.

dM
gM B
dt
(2)
A useful parallel description is a spinning toy top where the

axis of the top is analogous to M and gravity is analogous to
B. In the equilibrium state, the net magnetization M and
the static magnetic field B0 are parallel so that M does not
experience a torque and consequently the direction of M
does not change. Similarly, the axis of a spinning top
oriented vertically remains vertical.
The second magnetic field in MRI is an RF field that is
created using an RF amplifier that supplies oscillating
current into a coil that surrounds the patient. The coil is
designed to create a magnetic field, referred to as B1
field, oriented in the transverse plane and approximately
on the order of 10 T. By having the RFenergy oscillate at
the resonant frequency of the nuclei, this relatively low
field can perturb and rotate the net magnetization away
from its orientation along the longitudinal axis. The
resonant or Larmor frequency v0 is related to the static
field strength such that v0 gB0. For protons, the gyromagnetic ratio g/2p 42.57 MHz/T. The field created by
the tuned RF coil, referred to as an excitation, can be
Figure 2. (a) shows the precession of a spin with a magnetic

moment m in a static field with magnetic flux density of B0. An
assembly of spins in parallel and antiparallel states is shown in (b).
The Boltzmann equation determines the ratio of the spins in the
two states. As the components in x and y compensate each other,
the net magnetization M has a component in the z direction only
(parallel to B0). The coordinate system is shown with its unit
vectors i, j, and k along x, y, and z.
Z gradient coil
viewed as an applied torque that tips or flips spins away

from the longitudinal axis by an angle referred to as the
flip angle. The strength of the B1 field and the length of
time it is applied determine the flip angle. The flip angle
usually varies between 5 and 1808 depending on the
application.
Once the magnetization is no longer parallel to the static
field, the right-hand side of Equation 5 is no longer zero and
the direction of the net magnetization will change. In fact,
it will begin to precess about the axis of the static magnetic
field and at the Larmor frequency. In general, the precessional frequency is directly proportional to the magnetic
field experienced by the spin, such that v gB. Similar to a
toy top that is tipped an angle u off its vertical axis, the top
will maintain an angle of u as it rotates about the vertical
force supplied by gravity.
The net magnetization can be described by its longitudinal component Mz and its transverse component
Mxy, a complex value whose magnitude describes the components strength and whose angle describes the location of
the component in the xy plane. The rapid rotation of the
transverse component will create a time-varying magnetic
flux. A properly oriented receiver coil will detect this timevarying flux as a time-varying voltage, in agreement with
Faradays law of induction. Often, the same coil used for
excitation can also be used for reception. This voltage
signal, known as a Free Induction Decay, or FID, is shown
after a 908 excitation in Fig. 3, which is the most basic form
of a MR signal. Although the entire magnetization vector is
tipped into the transverse plane in this example, smaller
flip angles will also create a transverse magnetization and
thus an FID.
Figure 3. Generation of a free induction decay after a 908 RF

excitation.
The complex motion of the net magnetization, and thus

the recorded FID signal, can be described in a simplified
manner by using a rotating reference frame that rotates at
the Larmor frequency about the static B0 field. In this
rotating frame, the FID will decay as a simple exponential.
The causes of this decay, and its use as potential image
contrast mechanism, will be described after spatial encoding is described. Most MR signals are demodulating using
the Larmor frequency and, thus are effectively acquired in
the rotating frame.
Spatial Encoding
The first two magnetic fields described above allow
biological tissue to be polarized, perturbed, and measured.
In terms of clinical imaging, however, these fields merely
allow us to integrate the signal derived throughout the
body, a measure of little value. The field of MRI developed
only when a third spatially varying magnetic field, referred
to as a gradient field, was invented to spatially encode
the MRI signal. This method allows us to achieve submillimeter resolution while using RF energy whose wavelengths are on the order of tens of centimeters to meters.
Dr. Paul Lauterbur realized, in 1973, that, instead of
working like others to build a more homogenous field for
NMR spectroscopy, spatially varying the strength of the
magnetic field could provide a means to build an imaging
system. For this work, he won the Nobel Prize in Medicine
along with Sir Peter Mansfield in 2003.
The three gradient coils in a cylindrical MRI system, two
of which are shown in Fig. 1, are laid out concentrically on a
cylinder. The cylinder surrounds the patient as he or she lies
inside of the static B0 field. The three coils are designed to
create longitudinal magnetic fields in z that vary in strength
linearly with the x, y, and z dimensions, respectively. The
digital scanner hardware controls the current waveforms,
which are amplified by three respective gradient amplifiers
before being sent to the gradient coils. The strength of each
component gradient field, Gx,Gy, or Gz, is measured in G/cm
or mT/m and is directly proportional to the current supplied
to the coil. Changing gradient strengths quickly on clinical
scanners is possible with amplifiers capable of slew rates of
approximately 200 mT/m/s.
As the resonant frequency of an MR spin is directly
proportional to the magnetic field it experiences, a gradient
coil allows us to linearly vary the frequency of spins
according to their position within the magnet. For example,
a gradient of strength Gx, which does not vary in time,
causes the frequency of spins to vary linearly with the x
coordinate.
wx gB0 Gx x
Spins to the left of the magnet center rotate slower, spins

at the exact magnet center remain unchanged, and
spins to the right rotate faster than they did without
the gradient.
Gradients can be used to selectively excite only spins
from a slice or slab of tissue. Slice thicknesses in 2D MRI
range from 1 to 20 mm. To select a transverse slice, the z
gradient can be applied during RF excitation, which will
cause the resonant frequency to vary as a function of z in
285
the magnet, such that wz gB0 Gz z . Instead of exciting all the spins within the magnet, only spins whose
frequency matches the narrow bandwidth of a pulsedRF excitation will be excited within a slice at the center
of the magnet. Modulating the frequency of the RF pulse
up will move the slice superior in the body, whereas
modulating it down will excite an inferior slice. Likewise,
slices perpendicular to the x or y axis can be excited by
applying a Gx or Gy gradient, respectively, simultaneously
with RF excitation. In fact, an oblique slice orientation can
be achieved with a combination of two or more gradients.
The ability to control from which tissue signal is obtained,
without any patient movement, is a major advantage of
MRI.
Simplified MR Spatial Encoding. Once a slice of tissue is
selected, the two remaining spatial dimensions must be
encoded. A somewhat simplified method of visualizing
encoding follows. For a transverse slice, receiver data
could be obtained after RF excitation while a constant
gradient was applied in the x direction. Tuning a receiver
to select a very narrowband frequency range would determine which spins were present within a spatial range x1<
x< x1 Dx. By repeating the experiment while changing
the narrowband frequency range, a projection of the spin
densities along the x axis could be determined. Likewise,
the same experiment could be repeated while applying a
constant y gradient to obtain a projection of spin densities
along the y axis. Likewise, projections along arbitrary
axes could be achieved by acquiring data while applying
a combination of x and y gradients after RF excitation. In a
matter very similar to computed tomography (CT) imaging, an image could be reconstructed from this set of
acquired projections.
MR Spatial Encoding in the Fourier Domain. Although
possible, the proposed method would be very slow because
each sample point within each projection would require its
own MR experiment or excitation. Time between excitations in MR vary in duration from 2 ms to 4 s depending on
the desired image contrast. Even with the shortest excitation, each slice would require over 3 min of scan time. All
the data for an entire projection can be acquired within
milliseconds by considering how the phase of the transverse magnetization varies, instead of the frequency, with
spatial position. This description also uses the concept that
position in MR is encoding using an alternative Fourier
domain where signal location is mapped onto spatial frequencies.
Integrating the frequency expression in Equation 3
indicates how the spin phase, or location of the transverse
magnetization within the transverse plane, will vary with
the x coordinate during a general time-varying gradient
Gx(t) applied after excitation. Ignoring the phase term due
to the B0 field, which will be removed during demodulation
of the received signal, gives a phase term for each spin that
varies with the spatial position x and the integral of the
applied gradient at each point in time.
Rt
g
0
0
Mxy x; t Mxy xe j2 p2p t0 0 Gxt x dt
(4)
286
The signal received by the MR coil can be expressed as an

integration of all the excited spins in the xy plane using
the following equation:

Z Z
St
Mxy x; ye j2pkx tx dx dy
(5)
y
where a Fourier spatial frequency,

termed kx(t) in MR, has
g Rt
0
0
been substituted for 2p
t0 o Gx t xdt . In this example,
the coil simply integrates all the spins in the y dimension,
and thus only spatial information in x is available. The
received signal can be seen as a 1D Fourier transform of the
projection of transverse magnetization Mxy(x, y) onto the x
axis. The corresponding coordinate in the Fourier domain
at each point in time t is determined by the ongoing integral
of the gradient strength. Thus, we can acquire an entire
projection in one experiment rather than numerous MR
experiments as in the simplified example with the narrowband receiver.
The last spatial dimension for this 2D imaging example
y has a corresponding Fourier dimension termed ky, which
can be similarly traversed by designing the integral of the
Gy gradient current.
Z Z
Mxy x; te j2pkx tx e j2pky ty dxdy
(6)
St
y
g Rt
0
0
2p t0 o Gx t xdt .
where ky t
The integral of the gradients determines location in the Fourier space known as k
space in MRI. Numerous strategies can be used to traverse
and sample k space before transforming the data, often
with a Fast Fourier Transform (FFT), into the image
domain. The method can be extended to three dimensions
by exciting a slab of tissue and using the Gz gradient to
encode the third dimension.
As in the simplified example, a combination of Gx and Gy
can be used to sample the 1D Fourier expression of projections of Mxy at arbitrary angles. This data can be transformed into the actual projections using 1D inverse Fourier
transforms. Methods very similar to computed tomography
can translate the projection data into an image. Although acquiring data in this manner, known as radial
imaging, has interesting properties, by far the most popular method in clinical imaging traverses the Fourier space
in a Cartesian raster pattern known as spin-warp imaging.
This sampling is typically completed in a series
of experiments, where the time between consecutive
Figure 4. 2D spin-warp imaging with one

readout per TR. One complete line is
sampled along the readout direction on a
rectilinear grid in the Fourier domain (k
space) with a resolution of Dkx (circles on
the arrow). The next line is acquired
parallel at a distance of Dky on the grid by
increasing the phase-encoding gradient. This
scheme is repeated until the desired grid is
sampled. Images are reconstructed by an
inverse 2D Fourier transform (FT).
excitations is referred to as the repetition time TR. In

many MR acquisition schemes, a complete k space line is
acquired along kx, known as the frequency-encoding or
readout direction, for each TR. During each subsequent
TR, a line parallel to the previous one is sampled after
applying a short, pulsed Gy gradient. By changing the
strength of the pulsed Gy gradient by equal increments
during each MR experiment, a different phase shift is
placed on spins depending on their position in y. In terms
of the k space formalism, the area under the Gy gradient
pulse causes a vertical displacement in k space such that a
different horizontal line in k space is acquired in each MR
experiment, as shown in Fig. 4. Here, the vertical direction
in k space is known as the phase-encoding direction. By
applying 1D Fourier transforms in the kx direction, spin
position is resolved based on their frequency during readout. An image is formed by following these horizontal
transforms with 1D Fourier transforms in the ky dimension. Here, the y position of spins is resolved due to the
different phase shifts experienced in each experiment prior
to the readout gradient.
The image coverage, or field of view, in MRI decreases as
the sampling rate decreases. As MR samples in the frequency domain, failure to sample fast enough in k space
leads to aliasing in the image domain. Higher resolution in
MRI requires obtaining higher spatial frequencies or larger extents of k space. Achieving adequate resolution and
coverage then increases the amount of k space sampling
that is required and increases imaging time. Unlike other
modalities where hundreds to thousands of detectors can
be used at a time, encoding spatial position in this method
only allows one point of data to be taken at a time, which
explains MRs relatively slow acquisition speed. Industrial
scanners have only recently determined how to partially
bypass this limitation by using up to 32 different receivers
who have different spatial sensitivities to different tissues.
The differences of each receiver in proximity, and thus
sensitivity to each spin, can be used to synthesize unacquired regions of k space.
Image Contrast Through Varying Decay Rates
Imaging the spatial density distribution of hydrogen often
produces a very low contrast image, as the density of
hydrogen is relatively consistent in soft tissue. However,
the imaging experiments described above can be easily
modified to exploit the differences in time for which the
287
Figure 5. The regrowth of the longitudinal magnetization Mz (a) and the decay of the transverse
magnetization Mxy (b) after an RF excitation.
MR spins remain perturbed. The differences account for

the vast majority of image contrast in standard clinical
MRI.
After the spins are perturbed, the transverse magnetization decays toward zero, whereas, the longitudinal magnetization returns toward its equilibrium magnetization.
As more mechanisms exist for the loss of transverse magnetization than for the regrowth of longitudinal magnetization, the length of the magnetization vector M does not
remain constant after excitation. Although related, the
rate of longitudinal relaxation time, termed T1, is always
larger than the rate of transverse relaxation time, termed
T2.
If the magnetization has been completely tipped in the
transverse plane with a flip angle of 908, then the longitudinal magnetization recovers as
Mz M0 1 et=T1
(7)
T1 is also called the spinlattice relaxation time, because it

depends on the properties of the nucleus and its interactions with its local environment. The transverse relaxation
time T2 is also referred to as the spinspin relaxation
time, reflecting dephasing due to interactions between
neighboring nuclei.
Mxy Mxy 0et=T2
(8)
where Mxy(0) is the initial transverse magnetization

(Mxy(0) M0 for a 908 pulse). The temporal evolution of
the longitudinal and transverse magnetization is shown in
Fig. 5. In general, hydrogen protons in close proximity to
macromolecules have lower relaxation times than bulk
water that is freer to rotate and translate its position.
Delaying the encoding and acquisition of the transverse
magnetization until some time after RF excitation generates T2 image contrast. As injured and pathological tissues
generally have higher T2 relaxation rates, T2-weighted
images have positive image contrast. T1-weighting can
be achieved by using an interval between MR experiments,
the TR parameter, which does not allow enough time for
tissue to fully recover its longitudinal magnetization. Thus,
tissues with shorter T1 relaxation rates will recover more
quickly and thus have more signal present in the subsequent experiments used to build the image than tissues
with longer T1. In general, T1-weighting provides negative
contrast for pathological tissue. An example is shown in
Fig. 6 for a human brain tumor. The differing rates of
Figure 6. The flexibility of MR to image in different planes with different types of image contrast is
shown in these sagittal and axial brain tumor (arrows) images.
288
Table 1. Longitudinal (T1) and Transverse (T2) Proton

Nuclear Magnetic Resonance Relaxation Times for Several Tissues and Blood at 1.5 T
Tissue
T1/ms
T2/ms
Gray brain matter (3)

White brain matter (3)
Cerebrospinal fluid (CSF) (3)
Muscle (3)
Fatty tissue (3)
Oxygenated blood
De-oxygenated blood
950
600
4500
900
250
1200
1200
100
80
2200
50
60
220
120
recovery can also be used to null out an unwanted tissue,

such as fat, by inverting all the spins 1808 prior to imaging.
As the point where unwanted tissue passes through the
null of the recovery phase, an imaging experiment is
begun. This technique is referred to as inversion recovery
magnetization preparation or simply inversion recovery.
Table 1 lists representative relaxation times for some
tissues (3). Extensive reviews of the relaxations times (4)
and methods for their measurement (3,4) are available.
rotating faster than the Larmor frequency and others

spinning slower. The second pulse flips all magnetic
moments about an axis in the transverse plane, effectively
inverting the phase accruals due to different rotational
frequencies. Over the second interval td, the faster spins
will catch up with the slower spins. As a result, a spin echo
is said to form at the time 2td, also known as the echo time
or TE time. The amplitude of the signal at time TE is only

decreased due to T2 decay whereas the T2 effects have been
reversed. A simplified pulse sequence for the generation of
a spin echo is shown in Fig. 7b without the gradient
waveforms necessary for spatial encoding.
Signal-to-Noise Ratios
Signal in MR is generally proportional to the number of
nuclei and, thus, to the volume of the image voxel. Noise in
MR is caused by the random fluctuations of electrons in the
patient, and thus the source of noise is independent from
the signal generating sources. The data acquisition system
is designed such that the noise level from properly designed
MR electronics will be p
dominated
by patient noise. Overall,
SNR voxel volume total data sampling time.
Spin Echoes
Imaging Sequences
Ideally, the transverse magnetization decays according to

T2. However, the signal dephasing in the transverse plane
is significantly accelerated by field inhomogeneities due to
difference in magnetic susceptibility between tissue types
or the presence of paramagnetic iron. The largest inhomogeneities occur at air/tissue interfaces, such as near the
sinuses or near the diaphragm. These effects lead to different precession frequencies and loss of coherence that are
described by a T2 relaxation time
Ideally, all MRI would be performed with high spatial

resolution, a high signal-to-noise ratio (SNR), ultrashort
imaging time, and no artifacts. The difficulty in achieving
all of these properties simultaneously has led to the development of many acquisition methods that differ in image
contrast, acquisition speed, SNR, susceptibility to and type
of artifacts, energy deposited in the imaged patient, and
suppression of unwanted signal such as fat. Their corresponding images represent a combination of tissue-specific
parameters T1, T2, proton density r, and scan-specific
parameters such as repetition time (TR), echo time (TE),
flip angle, field of view (FOV), spatial resolution, and
magnetization preparation.
1
1
1
T2 T2 T20
(9)
where T0 2 represents the decay due to the effects described

above.
A method to reverse these often undesirable dephasing
effects uses a 908 pulse followed by a 1808 spin refocusing
pulse after a time delay td, as shown in Fig. 7b. The
first pulse rotates the longitudinal magnetization into
the transverse plane as in the case of the FID. Prior
to the second pulse, the magnetization dephases in the
transverse plane due to T2 effects, with some spins
Figure 7. Generation of a (a) free induction decay and (b) a spin

echo. In (a), other local factors dephase signal faster according to a
T2 decay rate. If a second RF pulse is applied at time td TE//2 to
flip the magnetization by 1808, the spins will refocus and form an
echo at TE 2td, which is only subject to T2 decay.
Gradient Recalled Echo (GRE) Imaging

Spin-echo imaging is desirable because signal voids due to
magnetic field inhomogeneity are avoided that could mask
pathological tissue or injury. Long repetition times, and
thus long scan times, are necessary in spin-echo imaging to
allow longitudinal magnetization to return after the relatively high flip angles used. Long scan times hinder the
capture of dynamic processes such as the beating heart,
cause discomfort to the patient, and limit the number of
patients who can be imaged with this expensive resource.
Thus, other methods of imaging have been developed. In
gradient recalled echo (GRE) imaging, the echo is formed
by dephasing and rephasing of the signal with gradient
fields as shown in Fig. 8. In these diagrams, known as pulse
sequence diagrams, plots of the time-varying gradient and
RF waveforms are shown as function of time. Compared
with the spin-echo sequences, gradient recalled echo imaging does not have a refocusing RF pulse. The absence of
this pulse allows for a reduced minimal repetition time and
echo time compared with spin-echo imaging, but the signal
becomes susceptible to T2* decay rather than T2 decay.
289
(SPGR) imaging. This technique is popular with contrastenhanced MR angiography, where an intravenously
injected paramagnetic contrast agent significantly
decreases the T1 of blood while the T1 of static tissues
remains unchanged.
In an opposite approach, known as steady-state free
precession (SSFP), the maximum amount of the transverse
magnetization is maintained by rewinding all gradients
prior to each RF pulse. The method provides T2-like contrast very quickly and has proven very popular when fast
imaging is essential such as in cardiac imaging.
Other Rapid MR Imaging Methods
Figure 8. Basic 2D gradient echo pulse sequence. First, the

magnetization is tipped into the transverse plane by an angle a
during the application of a slice select gradient Gz. Then, the
gradients Gy and Gx are used for phase encoding and the
readout gradient. An echo forms at t TE when the area under
the readout gradient is zero. This experiment is then repeated
every TR with a different phase encode.
After RF excitation, the signal is dephased along the readout direction x with a prewinding gradient lobe. The
amplitude of this gradient is then inverted to rephase
the spins. When the area under the readout gradient is
zero, the trajectory passes through the origin of k space and
the echo forms with maximum amplitude. During the
prewinder along the x axis, a gradient in y is played out
to produce y depending on phase shifts for phase encoding.
By using a flip angle less than 908, significant amounts
of transverse magnetization are available without the need
for long repetition times needed for recovering longitudinal
magnetization. For example, after a single 308 excitation,
the transverse magnetization contains sin(308) or one-half
of the available magnetization. Meanwhile, the longitudinal magnetization still contains cos(308) or nearly 87%
percent of the equilibrium magnetization. In fast GRE
imaging, the repetition time is significantly reduced and
generally less than the T2 values of biological tissues.
Under this condition, the transverse magnetization from
preceding RF pulses is not completely dephased and generally results in a complex superposition of echoes from
multiple RF pulses. Under certain conditions, a steadystate can be reached from repetition to repetition for one or
more components of the magnetization (6).
GRE sequences can be used to generate T1, T1/T2, T2,

T2 , and proton density-weighted contrast, depending on
the choice of TR, TE, the flip angle a, and the phase f of the
RF pulse. By altering the phase of the RF transmit pulse in
a pseudo-random method, the steady state of the transverse magnetization can be scrambled while the beneficial
aspects of the longitudinal steady state are maintained.
Although the signal from the transverse steady state is lost
and only the signal from the current RF pulse is available,
strongly T1-weighted images are available with this technique, known as RF spoiling or spoiled gradient recalled
In many applications, a short scan time is required to

reduce artifacts from physiological motion or to observe
dynamic processes. Many techniques exist to reduce the
scan time while preserving high spatial resolution. One
way to decrease spin-echo imaging time is to acquire multiple or all k space lines after a single preparation of the
magnetization as explored with RARE (Rapid Acquisition
with Relaxation Enhancement) (7). Also referred to as fast
or turbo spin echo, the method works by creating a train of
spin reversal echoes for which one line of k space is
acquired for each. In echo-planar imaging (EPI) (8), an
oscillating Gx gradient is used to quickly create many
gradient echoes. By adding small blip gradients in between
the negative and positive pulses of Gx, different horizontal
lines in k space can be acquired.
Although the first MRI method proposed the acquisition
of projections (9) as in CT, acquiring k space data on a
Cartesian grid is fairly robust to magnetic field inhomogeneities and other system imperfections. Although spinwarp imaging (10) is predominant today, k space can be
sampled along numerous 2D or 3D trajectories. The
PROPELLER technique (11) acquires concentric rectangular strips that rotate around the origin, as shown in
Fig. 8c. This method offers some valuable opportunities for
motion correction due to the oversampling of the center of k
space. K space can be sampled more efficiently with fewer
echoes using spiral trajectories (12), as shown in Fig. 8d.
Here, the amount of k space that can be acquired in one
excitation is limited only by T2 decay and possible blurring
due to off-resonance spins. In nonCartesian acquisitions,
phase errors due to off-resonance spins cause blurring. The
sampling trajectories for these acquisitions schemes are
shown in Fig. 9.
Applications
MRI is quickly moving beyond morphological and anatomical imaging. The advent of new functional image contrast mechanisms is making MR a tool for a much wider
group of people than radiologists. Psychology, psychiatry,
neurology, and cardiology are just some of the new areas
where MR is being applied. A description of application
areas in functional brain, diffusion-weighted brain, lung,
MR angiography, cardiac, breast, and musculoskeletal
imaging follows.
Functional Magnetic Resonance Imaging (fMRI). Functional Magnetic Resonance Imaging (fMRI) is a method of
290
Figure 9. 2D k space sampling trajectories. Shown are the spin-warp (a), radial sampling (b),
PROPELLER (c), and interleaved spiral imaging (d) examples.
measuring the flow of oxygenated blood in the brain (13

15). FMRI is based on the blood oxygen-level-dependent, or
BOLD, effect. BOLD MRI is accomplished by first exposing
a patient or volunteer to a stimulus or having them engage
in a cognitive activity while acquiring single-shot images of
their brain. The region of the brain that is responding to
the stimulus or is engaged in the activity will experience an
increase in metabolism. This metabolic increase will
require additional oxygen. Therefore, an increase in oxygenated blood flow will occur (oxyhemoglobin) to the local
brain area that is active. Oxyhemoglobin differs in its
magnetic properties from deoxyhemoglobin. Oxyhemoglobin is diamagnetic like water and cellular tissue. Deoxyhemoglobin is more paramagnetic than tissue, so it
produces a stronger MR interaction. These differences
between oxyhemoglobin and deoxyhemoglobin in BOLD
imaging are exploited by acquiring images during an
Figure 10. Color brain activation map is

superimposed on high resolution MR image. Signal
levels of the activated pixels are shown to increase
during cognitive activity periods, whereas they fall
off during periods of rest.
active state (more oxyhemoglobin) and in a resting

state (more deoxyhemoglobin), which creates a signal
increase in the active state and a signal decrease in
the resting state. Figure 10 shows a typical BOLD time
course (shown in black) where four active states and four
resting states exist. With prior knowledge of the activation timing (shown in red), we can perform a statistical test
on the data to determine which areas of the brain are
active. This statistical map (shown in color) is superimposed on a high resolution MR image so that one can
visualize the functional information in relation to relevant
anatomical landmarks.
Diffusion Imaging. The random motion of water molecules may cause the MRI signal intensity to decrease. The
NMR signal attenuation from molecular diffusion was first
observed more than a half century ago by Hahn (1950) (16).
Figure 11. Temporal schematic of a diffusion-weighted, spin-echo

pulse sequence with an EPI readout. The diffusion gradient pulses
are shown as gray boxes on the gradient axes. The direction of
diffusion-weighting can be changed by changing the relative
weights of the diffusion gradients along Gx, Gy, and Gz.
Subsequently Stejskal and Tanner (1965) described the

NMR signal attenuation in the presence of field gradients
(17). More recently, field gradient pulses have been used to
create diffusion-weighted MR images (18).
Diffusion-Weighted Pulse Sequences. Typically, diffusion-weighting is performed using two gradient pulses
with equal magnitude and duration on each side of a
1808 refocusing pulse, as shown in Fig. 11. The first gradient pulse dephases the magnetization as a function of
position, and the second pulse rephases the magnetization.
For stationary (e.g., no flow or diffusion) molecules, the
phases induced by both gradient pulses will completely
cancel, no signal attenuation will occur. In the case of
motion in the direction of the applied gradient, a net phase
difference will occur, Df gvGdD, which is proportional to
the velocity v, the area of the gradient pulses defined by the
amplitude G, and the duration d, and the spacing between
the pulses D. For the case of diffusion, the water molecules
are also moving, but in arbitrary directions and with
variable effective velocities. Thus, in the presence of diffusion gradients, the signal from each diffusing molecule will
accumulate a different amount of phase, which, after summing over a voxel, will cause signal attenuation. For simple
isotropic Gaussian diffusion, the signal attenuation for
the diffusion gradient pulses in Fig. 11 is described by
S So ebD where S is the diffusion-weighted signal, So is
the signal without any diffusion-weighting gradients
(but otherwise identical imaging parameters), D is the
apparent diffusion coefficient, and b is the diffusionweighting described by the properties of the pulse pair
b (gGd)2(D-d/3).
Diffusion Tensor Imaging. The diffusion of water in
fibrous tissues (e.g., white matter, nerves, and muscle) is
anisotropic, which means the diffusion properties change
as a function of direction. A convenient mathematical
model of anisotropic diffusion is using the diffusion tensor
(19), which uses a 3 3 matrix to describe diffusion using a
general 3D multivariate normal distribution. The diffusion
291
tensor matrix describes the magnitude, anisotropy, and

orientation of the diffusion distribution. In a diffusion
tensor imaging (DTI) experiment, six or more diffusionweighted images are acquired along noncollinear diffusion
gradient directions. Maps of the apparent diffusivity for
each encoding direction are calculated by comparing the
signal in an image without diffusion-weighting and the
signal with diffusion-weighting. The diffusion tensor may
then be estimated for each voxel, and maps of the mean
diffusion, anisotropy, and orientation may be constructed,
as shown in Fig. 12.
The primary clinical applications of diffusion-weighted
imaging and DTI are ischemic stroke (20,21) and mapping
the white matter anatomy relative to brain tumors and
other lesions (22). DTI is also highly sensitive to subtle
changes in tissue microstructure and, therefore, has
become a popular tool for investigating changes or differences in the microstructure as a function of brain development and aging, as well as disease.
Vascular Imaging. Magnetic Resonance Angiography
(MRA) describes a series of techniques that can be used to
image vascular morphology and provide quantitative blood
flow information in high detail. Two widely used techniques, phase contrast angiography and time-of-flight angiography, use the inherent properties of blood flow in the MR
environment to create angiograms. A third technique,
contrast-enhanced angiography, uses the injection of a
Figure 12. Representative diffusion tensor images. The images

are (top-left): a T2-weighted (or nondiffusion-weighted) image;
(bottom-left): a mean diffusivity map (note similar contrast to
T2-weighted image with cerebral spinal fluid appearing hyperintense); (top-right): a fractional anisotropy map (hyperintense in
white matter); and (bottom-right) the major eigenvector direction
indicated by color (red R/L, green A/P, blue S/I) weighted by
the anisotropy (note that specific tract groups can be readily identified).
292
paramagnetic contrast agent into the vascular system to

specifically alter the magnetic properties of the blood in
relation to the surrounding tissue.
Phase-contrast (PC) angiography (23) usually uses a
pair of gradient pulses of equal strength and opposite
polarity, placed in the MRI sequence between the RF
excitation pulse and the data acquisition window. During
the imaging sequence, stationary nuclei accumulate phase
during the first gradient pulse, and accumulate the opposite phase during the second gradient pulse, resulting in
zero net phase. Moving nuclei accumulate phase during the
first gradient pulse, but during the second pulse are in
different positions, and accumulate phase different from
that obtained during the first pulse. The net accumulated
phase is proportional to the strength of the gradient pulses
and the velocity of the nuclei. From the resulting data,
images can be formed of both blood vessel morphology and
blood flow.
TOF angiography techniques (24) (more accurately
called inflow techniques) typically use a conventional
gradient-echo sequence to acquire a thin 3D volume or a
series of 2D slices. The nuclei in stationary tissue are
excited by many consecutive slice-selective RF pulses. As
a short TR is used, the longitudinal magnetization is not
able to return to equilibrium, resulting in saturation of
magnetization and low signal. Moving nuclei in the blood
flow into the slice during each TR period, having been
excited by zero or very few RF pulses. As these nuclei
arrive in the imaging slice at or near full equilibrium
magnetization, high signal is obtained from blood.
Figure 13 shows a projection image of a 3D TOF dataset
acquired in the head. The TOF technique can produce high
Figure 13. Time-of-flight (TOF) angiography in the head uses

inflow of fresh blood to produce contrast between blood and the
surrounding tissue. A Maximum Intensity Projection (MIP)
reformatted image is used to compress the acquired volume
data into a single slice for display.
quality MRA images in many situations, but slow or inplane blood flow can result in blood signal saturation and
reduced the quality of the images.
Contrast-enhanced MRA (CE-MRA) is performed using
an injection of a paramagnetic contrast agent into the
intravenous bloodstream (25). Although several transition
and rare-earth metal ions can be used, the most common is
Gadolinium (Gd+3) chelated to a biologically compatible
molecule. The compound is paramagnetic, having a strong
dipole moment and generating strong local magnetic field
perturbations, which increase the transfer of energy
between the excited hydrogen nuclei and the lattice, promoting T1 relaxation and return to equilibrium of the
longitudinal magnetization.
The contrast agent is injected intravenously in a limb
away from the area of interest and circulates into the
arterial system. The longitudinal relaxation rate is typically enhanced by a factor of 15 to 25 during this initial
arterial phase, resulting in a much shorter T1 for blood
compared with the surrounding tissue. As the longitudinal
magnetization in blood is much higher after each TR
period, background tissue is suppressed in a manner similar to TOF imaging, and blood vessels have a comparably
bright signal on the resulting images. Imaging is typically
performed so that the central k space lines, which contain
most of the image contrast information, are acquired while
the contrast agent is distributed in the arteries of interest,
but before it can circulate into the neighboring veins.
MRA data consist of large volumetric sets of image data,
which are stored in the format of contiguous image slices.
Specialized image display techniques are used to display
the data in a manner that can be interpreted by the
radiologist. Maximum Intensity Pixel (MIP) projections
are widely used and are formed by projecting the volume
set of data onto a single image plane. Here, each image
pixel is obtained as the maximum value along the corresponding projection, as shown in Fig. 13. Volume rendering
is beginning to be used more often to display MR angiograms. The individual slices of data are always available
for detailed review by the radiologist and can be reformatted into any plane on the computer workstation to
optimally display the vasculature of interest.
Cardiac MRI. Cardiac magnetic resonance (CMR) imaging is an evolving technique with the unprecedented
ability to depict both detailed anatomy and detailed function of the myocardium with high spatial and temporal
resolution. The past decade has seen tremendous development of phased array coil technology, ultra-fast imaging
sequences, and parallel imaging techniques, all of which
have facilitated ultra-fast imaging methods capable of
capturing cardiac motion during breath-holding. The ability to perform imaging in arbitrary oblique planes, the lack
of ionizing radiation, and the excellent soft tissue contrast
of MR make it an ideal method for cardiac imaging.
Comprehensive cardiac imaging is performed routinely
in both in-patient and out-patient settings across the
country and is widely considered the gold standard for
clinical evaluation of many cardiac diseases (26).
Ischemic heart disease caused by atherosclerotic coronary artery disease (CAD) is the leading cause of mortality,
morbidity, and disability in the United States, with over 7

million myocardial infarctions and 1 million deaths every
year (27). Consequently, ischemic heart disease is the
primary indication for CMR. Accurate visualization of wall
thickness and global function (ejection fraction), as well as
focal wall motion abnormalities, is performed with retrospectively ECG-gated ultra-fast short TR pulse sequences,
especially steady-state gradient recalled echo imaging (28),
as shown in Fig. 14. Breath-held cinemagraphic or CINE
images have high SNR, excellent blood to myocardial
contrast, and excellent temporal resolution (< 4050 ms)
capable of detecting subtle wall motion abnormalities.
Areas of myocardial infarction (nonviable tissue) are exquisitely depicted with inversion recovery (IR) RF-spoiled
gradient echo imaging, acquired 1020 minutes after intravenous injection of gadolinium contrast (29), as shown in
293
Fig. 14. Areas of normal myocardium appear dark, whereas

regions of nonviable myocardium appear bright (delayed
hyper-enhancement). Accurate depiction of subtle myocardial infarction is possible because of good spatial resolution
across the heart wall. The combination of motion and
viability imaging is a powerful combination. Areas with
wall motion abnormalities but without delayed hyperenhancement may be injured or under-perfused from a
critical coronary artery stenosis but are viable and may
benefit from revascularization.
Cardiac stress testing using CMR has seen increasing
use for the evaluation of hemodynamically significant
coronary artery stenoses (30). Imaging of the heart during
the first pass of a contrast bolus injection using rapid
T1-weighted RF-spoiled gradient echo sequences is a
highly sensitive method for the detection of alterations
Figure 14. End-diastolic (a), mid-systolic (b), and end-systolic (c) short axis CINE images of the
heart in a patient with a myocardial infarction in the anterior wall and septum, demonstrated by
decreased wall thickening (arrows in b, c). The corresponding T1-weighted RF-spoiled gradient
recalled echo first-pass perfusion image (d) shows a fixed perfusion deficit (darker myocardium) in
the corresponding territory (arrows). Finally, an inversion recovery RF-spoiled gradient echo image
acquired at the same location (e) demonstrates a large region of delayed hyper-enhancement
(arrows) indicating a full wall thickness region of nonviable myocardium that corresponds to the
region of decreased perfusion and decreased contraction.
294
in myocardial blood flow (perfusion). Perfusion imaging

during both stress (pharmacologically induced) and rest
can reveal reversible perfusion defects that reflect a
relative lack of perfusion during stress. In this way,
coronary reserve can be evaluated and CAD can be
uncovered, leading to further evaluation with coronary
catheterization and possible angioplasty and stenting.
Direct imaging of coronary arteries with CMR has shown
tremendous technical advances, but is not commonly
used, except for imaging of proximal coronary arteries
in the evaluation of anomalous coronary arteries.
Other important indications of CMR include congenital
heart disease, primarily, but not exclusively, in the
pediatric population (31). Accurate diagnosis of a wide
variety of congenital abnormalities requires high resolution, high contrast imaging that permits depiction of complex anatomical variants seen with congenital heart
disease. Although anatomic imaging can be performed
accurately with cardiac-gated CINE sequences, conventional sequences such as cardiac-gated black-blood fast
spin-echo (FSE) and T1-weighted spin-echo imaging are
invaluable tools. Equally important to accurate anatomical
imaging is functional imaging. With altered anatomy
comes radically altered hemodynamics, requiring visualization of myocardial function with CINE imaging. Phasecontrast velocity imaging permits flow quantification
through the heart, including the great vessels (pulmonary
artery, aorta, etc.). An important example includes quantification of left-to-right shunts with resulting over-circulation of the pulmonary circulation. With a wide variety
of pulse sequences, flexible scan plane prescription and the
lack of ionizing radiation, CMR is ideally suited for evaluation of congenital heart disease.
Other important applications of CMR include visualization of valvular disease, pericardial disease, valvular disease, and cardiac masses. The latter two are particularly
well evaluated with CMR; however, they are relatively
uncommon and will not be discussed here.
Hyperpolarized Contrast Agents in MRI. Conventional
MR imaging measures the resonant signal from the hydrogen nuclei of water, the most ubiquitous and highly concentrated component of the body. However, many other
nuclei exist with magnetic dipole moments that produce
MR signals. Many of these nuclei, such as phosphorous-31
and sodium-23, are biologically important in disease processes. However, these species typically exist at a very low
concentration in the body, making them difficult to image
with sufficient signal. One approach is to align, or polarize,
the nuclei preferentially using physical processes other
than the intrinsic magnetic field of the MR scanner. In
some cases, these polarization processes can align many
more nuclei than otherwise possible. These hyperpolarized
nuclei can then act as contrast agents to better visualize
blood vessels or lung airways on MRI. For example,
helium-3 and xenon-129 are inert gases whose magnetic
dipole moments can be hyperpolarized using spinexchange optical pumpinga method of generating a preferred alignment of the nuclear dipoles using polarized
laser light (32). As they are inert gases, polarized
helium-3 and xenon-129 are used as inhaled contrast
agents for visualizing the lung airspaces (upper-right

panel) using MRI (33),(34). Unlike other parts of the body,
conventional MRI of the lungs suffers from poor signal due
to low water proton density and the multiple air-tissue
interfaces that further degrade the MR signal in the upper
left of Fig. 15. Hyperpolarized gas MRI has been particularly useful for depicting airway obstruction in several lung
diseases including asthma (lower panel of Fig. 15) (35),
emphysema (36), and cystic fibrosis (37). Additional techniques based on this technology show promise for MR
imaging of blood vessels using injected xenon-129 dissolved
in lipid emulsion (38), gas-filled microvesicles (39) and
liquid-polarized carbon-13 (40). Hyperpolarized carbon13 agents are of particular interest because of the wide
range of biologically active carbon compounds in the body.
Another important advantage of this technology is its
ability to maintain high signal using low magnetic field
(0.10.5 T) scanners (41). These systems are much cheaper
to purchase and maintain than the high field (1.53.0 T)
MRI scanners in common clinical use today.
Breast MRI. Breast MRI is presently used as an adjunct
to mammography and ultrasound for the detection and
diagnosis of breast cancer. Dynamic contrast-enhanced
(DCE) MRI has been shown to have high sensitivity
(83%96%) to breast cancer but has also demonstrated
variable levels of specificity (3789%) (42). DCE-MRI
requires an injection of a contrast agent and acquisition
of a subsequent series of images to enable the analysis of
the time course of contrast uptake in suspect lesions.
Lesion morphology is also important in discerning benign
from malignant lesions in breast MRI. Standard in-plane
spatial resolution is sub-millimeter. A typical clinical
breast MRI includes a spoiled gradient echo (SPGR) T1weighted sequence both precontrast (Fig. 16a) and, at
minimum, at 30 second intervals postcontrast (Fig. 16b).
Along with their morphologic characteristics, lesions can
be further described by the shape of their contrast uptake
curve. The three general categories of contrast uptake are
(1) slow, constant contrast uptake (2) rapid uptake and
subsequent plateau of contrast, and (3) rapid uptake and
rapid washout of contrast (43). Although the slowly enhancing lesions are usually benign and fast uptake and washout is a strong indication of malignancy, time course lesion
characterization is not absolute. The ambiguity of time
course data for certain classes of lesions drives the investigation into higher temporal resolution imaging methods.
A standard clinical breast MRI also includes acquisition of
a T2-weighted sequence for the identification of cysts
(Fig. 16c). Present research in breast DCE-MRI is focused
on development and application of pulse sequences that
provide high temporal and spatial resolution. Also, investigation is ongoing into more specific characterization of
uptake curves. Diffusion-weighted MRI, blood-oxygenlevel-dependent imaging, and spectroscopy are also being
investigated as possible methods to improve the specificity
of DCE-MRI in the breast. In some circumstances, the high
sensitivity of breast DCE-MRI outweighs the variable
specificity leading to the present use of DCE-MRI to determine the extent of disease, with equivocal mammographic
findings, and for the screening of high risk women.
Figure 15. Upper left shows normal lack of signal in parenchyma of lungs in MRI. Ventilated areas
are clearly seen after imaging inhaled hyperpolarized helium. Rapid imaging during inhalation and
exhalation shows promise for capturing dynamic breathing processes.
Figure 16. Fat-suppressed (a) pre-contrast T1-weighted image (b) postcontrast T1-weighted image
(c), and noncontrast T2- weighted image. Images are from different patients.
295
296
MRI of Musculoskeletal Disease. Musculoskeletal imaging

studies traumatic injury, degenerative changes, tumors,
and inflammatory conditions of the bones, tendons, ligaments, muscles, cartilage, and other structures of joints.
Although X-ray imaging is the work-horse imaging modality for many musculoskeletal diseases, MRI plays a
critical role in several aspects of diagnosis, staging, and
treatment monitoring.
Fast spin-echo (FSE) pulse sequences are typically
used to acquire T1, T2, and proton density-weighted
images of the joints. For the assessment of joint structures,
images are acquired in multiple planes to ensure adequate
spatial resolution in all dimensions. These MR images
can be used to evaluate tissues including ligaments,
bone, cartilage, meniscus, and labrum. As a result of
their high spatial resolution and excellent soft tissue
contrast, MR images can provide accurate diagnosis that
can prevent unnecessary surgeries and can facilitate
pre-operative planning when surgical intervention is
required.
Osteoarthritis is a degenerative disease that
affects approximately 20 million Americans and countless
others around the world. Currently, this debilitating disease is often not detected until the patient experiences
pain that can be reflective of morphologic changes to joint
cartilage. MR can be used to accurately measure cartilage
Figure 17. Axial knee image in a

patient with inflamed synovium
shows excellent soft tissue contrast
available in MRI. The thickened
intermediate signal intensity for
synovium (arrowheads) is well distinguished from the adjacent high
signal intensity of joint fluid (arrow).
thickness and volume. New MR techniques are also

under development that may provide insight into biochemical changes in cartilage at the earlier stages of
osteoarthritis that precede gross morphologic changes
and patient pain.
Fortunately, primary bone tumors are relatively rare.
However, bone is a common site for metastatic disease,
which is especially true for breast, lung, prostate, kidney,
and thyroid cancers. MR is a sensitive test for metastatic
bone disease and is being adopted as a standard of care in
some parts of the world, replacing nuclear scintigraphy. A
typical approach employs inversion recovery pulse
sequences to generate fat-suppressed, T2-weighted images.
Diffusion-weighted imaging also shows promise to detect
hemotalogic cancers such as multiple myeloma, leukemia,
and lymphoma.
Inflammatory diseases include infection and inflammatory forms of arthritis. Infection of the foot is a common
complication of microvascular disease often seen with diabetes, a disease afflicting 18 million Americans. MR can be
used to assess the vasculature of the foot as well as diagnose infection and evaluate treatment efficacy. Two million
Americans have rheumatoid arthritis, a common form of
inflammatory arthritis. Inflammation from a condition
known as synovitis, which often occurs in rheumatoid
arthritis patients, is shown in Fig. 17. New MR methods
are being developed to detect rheumatoid arthritis earlier

and to gauge treatment success.
BIBLIOGRAPHY
1. Liang Z-P. Lauterbur PC. In: Akay M, ed. Principles of
Magnetic Resonance ImagingA Signal Processing Perspective. IEEE Press Series in Biomedical Engineering. New
York: IEEE Press; 2000. p 416.
2. Abragam A. Principles of Nuclear Magnetism. Oxford, UK:
Oxford University Press; 1994.
3. Haacke EM, et al. Magnetic Resonance Imaging-Physical
Principles and Sequence Design. New York: Wiley; 1999. p 914.
4. Bottomley PA, et al. A review of H-1 nuclear-magneticresonance relaxation in pathologyare T1 and T2 diagnostic.
Med Phys 1987;14(1):137.
5. Kingsley PB. Methods of measuring spin-lattice (T-1) relaxation times: An annotated bibliography. Concepts Magn Reson
1999;11(4):243276.
6. Scheffler K. A pictorial description of steady-states in rapid
magnetic resonance imaging. Concepts Magn Reson 1999;
11(5):291304.
7. Hennig J, Nauerth A, Friedburg H. RARE imaging: A fast
imaging method for clinical MR. Magn Reson Med 1986;
3(6):823833.
8. Mansfield P. Multi-planar image formation using NMR spin
echoes. J Phys C: Solid State Phys 1977;10:L55L58.
9. Lauterbur PC. Image formations by induced local interactions: Examples employing nuclear magnetic resonance. Nature 1973;242:190191.
10. Edelstein WA, et al. Spin warp NMR imaging and applications to human whole-body imaging. Phys Med Biol
1980;25(4):751756.
11. Pipe JG. Motion correction with PROPELLER MRI: Application to head motion and free-breathing cardiac imaging.
Magn Reson Med 1999;42(5):963969.
12. Meyer CH. et al. Fast spiral coronary artery imaging. Magn
Reson Med 1992;28(2):202213.
13. Ogawa S, Lee T-M, Nayak A, Glynn P. Oxygenation-sensitive
contrast in magnetic resonance image of rodent brain at high
magnetic fields. Magn Reson Med 1990;14:6878.
14. Ogawa S, et al. Brain magnetic resonance imaging with
contrast dependent on blood oxygenation. Proc Natl Acad
Sci USA 1990;87:98689872.
15. Bandettini PA, et al. Time course EPI of human brain function during task activation. Magn Reson Med 1992;25(2):
390397.
16. Hahn E. Spin echoes. Phys Rev 1950;80(4):580594.
17. Stejskal E, Tanner J. Spin diffusion measurements: Spin
echoes in the presence of a time-dependent field gradient.
J Chem Phys 1965;42(1):288292.
18. Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E,
Laval-Jeantet M. MR imaging of intravoxel incoherent
motions: Application to diffusion and perfusion in neurologic
disorders. Radiology 1986: 161(2):401407.
19. Basser PJ, Mattiello J, LeBihan D. MR diffusion tensor
spectroscopy and imaging. Biophys J 1994;66:259267.
20. Moseley ME, Kucharczyk J, Mintorovitch J, Cohen Y, Kurhanewicz J, Derugin N, Asgari H, Norman D. Diffusionweighted MR imaging of acute stroke: Correlation with T2weighted and magnetic susceptibility-enhanced MR imaging
in cats. AJNR Am J Neuroradiol 1990;11(3):423429.
21. Warach S, Dashe JF, Edelman RR. Clinical outcome in
ischemic stroke predicted by early diffusion-weighted and
perfusion magnetic resonance imaging: A preliminary analysis. J Cereb Blood Flow Metab 1996;16(1):5359.
297
22. Witwer BP, Moftakhar R, Hasan KM, Deshmukh P, Haughton V, Field A, Arfanakis K, Noyes J, Moritz CH, Meyerand
ME, Rowley HA, Alexander AL, Badie B. Diffusion-tensor
imaging of white matter tracts in patients with cerebral
neoplasm. J Neurosurg 2002;97(3):568575.
23. Dumoulin CL, Hart HR. Magnetic resonance angiography.
Radiology 1986;161:717720.
24. Keller PJ, Drayer BP, Fram EK, Williams KD, Dumoulin CL,
Souza SP. MR angiography with two-dimensional acquisition
and three-dimensional display. Radiology 1989;173:527
532.
25. Prince MR, Yucel EK, Kaufman JA, Harrison DC, Geller
SC. Dynamic gadolinium-enhanced three-dimensional abdominal MR arteriography. J Magn Reson Imaging 1993;3:
877881.
26. Gibbons RJ, Araoz PA. The year in cardiac imaging. J Am
Coll Cardiol 2005;46(3):542551.
27. Association AH, ed. 2005: Heart Disease and Stroke Statistics
2005 Update. Dallas, TX: A.H. Association; 2005.
28. Carr JC, Simonetti O, Bundy J, Li D, Pereles S, Finn JP. Cine
MR angiography of the heart with segmented true fast imaging
with steady-state precession. Radiology 2001;219(3):828834.
29. Kim RJ, Wu E, Rafael A, Chen EL, Parker MA, Simonetti O,
Klocke FJ, Bonow RO, Judd RM. The use of contrastenhanced magnetic resonance imaging to identify reversible
myocardial dysfunction. N Engl J Med 2000;343(20):1445
1453.
30. Ray T. Magnetic resonance imaging in the assessment of
coronary artery disease. Curr Atheroscler Rep 2005;7(2):
108114.
31. Rickers C, Kraitchman D, Fischer G, Kramer HH, Wilke N,
Jerosch-Herold M, et al. Cardiovascular interventional MR
imaging: A new road for therapy and repair in the heart.
Magn Reson Imaging Clin N Am 2005;13(3):465479.
32. Bouchiat M, Carver T, Varnum C. Nuclear polarization in
3He gas induced by optical pumping and dipolar exchange.
Phys Rev Lett 1960;5:373375.
33. Albert MS, Cates GD, Driehuys B, et al. Biological magnetic
resonance imaging using laser-polarized 129Xe. Nature
1994;370:199201.
34. van Beek E, Wild J, Kauczor H, Schreiber W, Mugler J, Lange
E. Functional MRI of the lung using hyperpolarized 3-helium
gas. J Mag Reson Imag 2004;20:540554.
35. Samee S, Altes T, Powers P, et al. Imaging the lungs in
asthmatic patients by using hyperpolarized helium-3 magnetic resonance: Assessment of response to methacholine and
exercise challenge. J Allergy Clin Immunol 2003;111: 1205
1211.
36. Salerno M, Lange E, Altes T, Truwit J, Brookeman J, Mugler
J. Emphysema: Hyperpolarized helium 3 diffusion MR imaging of the lungs compared with spirometric indexesinitial
experience. Radiology 2002;222:252260.
37. Altes TA, de Lange EE. Applications of hyperpolarized
helium-3 gas magnetic resonance imaging in pediatric lung
disease. Top Magn Reson Imaging 2003;14:231236.
38. Moller HE, Chawla MS, Chen XJ, et al. Magnetic resonance
angiography with hyperpolarized 129Xe dissolved in a lipid
emulsion. Magn Reson Med 1999;41:10581064.
39. Callot V, Canet E, Brochot J, et al. MR perfusion imaging
using encapsulated laser-polarized 3He. Magn Reson Med
2001;46:535540.
40. Goldman M, Johannesson H, Axelsson O, Karlsson M. Hyperpolarization of 13C through order transfer from parahydrogen: A new contrast agent for MRI. Magn Reson Imag
2005;23:153157.
41. Parra-Robles J, Cross AR, Santyr GE. Passive shimming of
the fringe field of a superconducting magnet for ultra-low
298
MAMMOGRAPHY
field hyperpolarized noble gas MRI. J Magn Reson 2005;174:

116124.
42. Kvistad KA, et al. Breast lesions: Evaluation with dynamic
contrast-enhanced T1-weighted MR imaging and with T2weighted first-pass perfusion MR imaging. Radiology 2000;
216:545553.
43. Kuhl CK, Schild HH. Dynamic interpretation of MRI of the
breast. JMRI 2000;12:965974.
MAGNETOCARDIOGRAPHY. See BIOMAGNETISM.

MANOMETRY, ANORECTAL. See ANORECTAL
MANOMETRY.
MANOMETRY, ESOPHAGEAL.
See ESOPHAGEAL
MANOMETRY.
MAMMOGRAPHY
PEI-JAN PAUL LIN
Beth Israel Deaconess
Medical Center
Boston, Massachusetts
INTRODUCTION
It has been shown that reduced breast cancer mortality
in the past decade can be attributed to the high sensitivity
of screening mammography in detecting nonpalpable
lesions (1,2). There has been a wide variety of equipment
and imaging modalities employed in the breast cancer
detection and imaging; ranging from ultrasound imager,
X-ray mammography, computed tomography scanner (CT),
to magnetic resonance imager (MRI). Additionally, thermography, light diaphanography, electron radiography,
and microwave radiometry have also been utilized, experimentally, to detect breast cancer without much success.
Brief explanations of these imaging modalities have been
described in the first edition of this Encyclopedia, NCRP
Report No.85, and in a review article by Jones (35).
Among those modalities; ultrasound, X-ray mammography, CT, and MRI, X-ray mammography is the most practical and relatively inexpensive, and is the only main
stream of equipment available for breast cancer detection.
At present, the ultrasound imager is often employed as an
adjunct to the X-ray mammography and is not the primary
screening imaging device. However, when mammography
is mentioned, it is normally meant to say X-ray mammography. For these reasons, this article will devote all of its
effort to X-ray mammography.
detection of breast cancer in its earliest, most treatable

stages. Thus, it is required by law that facilities providing
mammography services be properly accredited and be
certified by the U.S. Food and Drug Administration
(FDA). Accreditation and Certification of mammography facilities are beyond the scope of this article, and
interested readers are requested to refer to the FDAs
WEB Site (6), and the American College of Radiology
(ACR) WEB Site (7).
X-RAY MAMMOGRAPHY
Conventional X-ray equipment was initially employed for
breast cancer imaging with industrial (thick) emulsion film,
or portal imaging film for use in radiation therapy as the
image receptor in order to visualize the small microcalcifications, prior to, as late as 1970s. The breast entrance dose of
this imaging process exceeded well over 85 mGy per film
(10 R per film) and the radiographic techniques were
typically in the range of 4555 kVp, and 1000 mAS with
a radiation beam quality of half-value layer (HVL) 1.0
1.5 mm of aluminum (mmAl) (8). The X-ray beam spectrum
produced by a conventional X-ray tube, equipped with tungsten anode, is not necessarily optimized for breast cancer
detection. The mammography images obtained in this manner had the desired spatial resolution (20 lp mm1), but
had a less than desirable radiographic contrast. This combination of high entrance dose with low radiographic
contrast was not an acceptable approach. The radiology
community was searching for a new breast imaging solution.
In the 1970s, xeromammography imaging plates provided
the much needed improvement in image quality and lowered the breast entrance dose by a factor of two thirds to
one half compared to using the thick emulsion industrial
type film (9).
Due to its unique patient positioning of breast imaging,
the geometrical arrangement of dedicated mammography
units should be pointed out. As shown in Fig. 1, with the
X-ray tube
Collimator
Radiaion field
Patient
Examination table
THE MAMMOGRAPHY QUALITY STANDARDS ACT

OF 1992 (MQSA), (PUBLIC LAW 102539)
Breast cancer was a major public health issue in the
early 1990s; U.S. Congress enacted MQSA to ensure that
all women have access to quality mammography for the
Screen-film cassette
Anti-scatter grid
Figure 1. Geometric arrangement of conventional radiography.
The X-ray system, the anatomy of interest, and the screen-film
cassette are centered and aligned for exposure.
MAMMOGRAPHY
Lateral View of Mammography Examination Setup
X-ray tube
Collimator
Radiaion field
Chest wall
Compression
paddle
Breast holder
Anti-scatter grid
Screen-film cassette
(a) Correct alignment of X-ray
system with the patient's
chest wall included in the
field-of-view.
(b) Improper setup. The "black"

triangular shaped area is not
in the field-of-view.
Figure 2. Geometry of dedicated mammography systems. On the

left side of (a), the geometrical arrangement of a dedicated
mammography unit is correctly setup where the center of the
radiation beam is aligned to the chest wall of the patient with
the compression cone pressing down on the breast being imaged.
On the right side (b), improper setup for mammography
examination is evident. The black triangular area is not
included in the image captured by the image receptor,
potentially missing the suspected cancer site.
conventional radiography system, the X-ray tube is setup

in the center of the anatomy of interest. The radiation field
is a projection of diverging X rays restricted by the collimator, centered to the area of interest. The geometry of a
dedicated mammography system on the other hand is offset so as to maximize inclusion of breast tissue close to the
chest wall (see Fig. 2a). If the geometry were setup in the
same manner as the conventional radiography as depicted
in Fig. 2b, the black triangular area of the breast would not
be included in the imaging field possibly missing a suspected cancer site.
THE PHYSICS OF MAMMOGRAPHIC IMAGING

Considering that various tissues in the breast are radiologically similar, but not identical, the task of differentiating the fibrous, ductal, and glandular tissues is extremely
difficult. This is due to the fact that (1) water; the main
ingredient of human tissue, has a density of r 1.0 g cm3
and effective atomic number of Zeff 7.47.6, and (2) fat
has a density of 0.9 g cm3 and effective atomic number
of Zeff 5.96.5 (10). Thus, various breast tissues with
varying degrees of fat and water contents are very similar
from radiological point of view.
The development of screen-film mammography was, in
reality, paired with the redesigning of dedicated X-ray
equipment for breast imaging. Breasts are relatively thin
(physical thickness and X-ray attenuation property) compared to other body parts. Radiographs of extremities, for
example, are obtained with X-ray tube potential in the
299
range of 5060 kVp. Breast tissues contain no high

attenuation anatomy, such as bones, a lower tube potential
(< 35 kVp) would be more suitable for breast imaging (11).
Use of lower tube potential has a potential benefit of taking
advantage of the photoelectric effect in differentiating the
subtle differences of breast tissues.
It should be pointed out that there are five basic ways
that X-ray photons interact with matter; they are (1)
coherent scattering, (2) photoelectric effect, (3) Compton
effect, (4) pair production, and (5) photodisintegration
(12). Of these five interactions, photoelectric effect and
Compton effect predominantly contribute to the image
formation in diagnostic radiology. The probabilities of
photoelectric effect and the Compton effect can be
expressed as following;
Photoelectric effect Z3 =E3
Compton effect E Z
In equations (1) and (2), E is the photon energy, and Z is

the atomic number. Clearly, from Eq. 1, the lower the Xray photon energy, and the higher the atomic number the
probability of the photoelectric effect will be higher. Since
the photoelectric effect is proportional to the cube of
(Z/E), the differential of breast tissues in Zeff is enhanced in the image formation. Thus, a better approach
to differentiate and image the breast tissues is to employ a
lower X-ray tube potential for imaging. While this
approach is good for imaging, the radiation dose
absorbed by the breast would still be a major concern.
Conventional X-ray tubes employ tungsten as the target
material and produce a broad X-ray spectrum through the
bremsstrahlung process. At X-ray tube potential of 35 kVp
and lower, the tube potential used on most dedicated
mammography systems, the tungsten target X-ray tubes
would produce broad energy spectrum that contribute less
to the image formation and more to the radiation dose.
Tungsten, rhodium, and molybdenum are ideal for use in
X-ray production due to their relatively high melting points
than other metallic elements. Typically, dedicated mammography equipment is equipped with molybdenum target
X-ray tube with beryllium window (port), and 30-mm thick
molybdenum filter. The X rays, generated at 2530 kVp
tube potential, produced by the molybdenum target X-ray
tube contain a large fraction of characteristic X rays at
energies 1720 keV (13,14). It is, therefore, quite natural
to optimize and utilize these characteristic X rays for image
formation and, consequently, patient exposure reduction at
the same time. Most dedicated mammography equipment
employ a combination of molybdenum target with aluminum, molybdenum, or rhodium filters. The characteristic
X rays generated at the molybdenum target have X-ray
photon energies of 17.4 and 19.6 keV, just below the
K-absorption edge of 20 keV, thus are quite transparent
to the molybdenum filter. This is illustrated in Fig. 3. On
the left of Fig. 3a, is a schematic drawing of the X-ray
spectrum generated at 30 kVp with a molybdenum target
and aluminum filter. In the middle of Fig. 3b, is the same
system with 30 mm thick molybdenum filter. Notice that
the K-absorption edge curve (dashed curve) of molybdenum
shows that the X-ray photons with energies just above
MAMMOGRAPHY
(a) Mo Target/Al Filter
(b) Mo Target/Mo Filter
K=17.4
0.8
0.6
K =19.6
0.4
0.2
0
(c) Rh Target/Rh Filter
1.0
Relative Intensity
Relative Intensity
1.0
10
20
30
Photon Energy (keV)
0.8
1.0
K-edge=20.0
0.6
0.4
0.2
0
10
20
30
Photon Energy (keV)
Relative Intensity
300
0.8
K=20.2
K=22.7
K-edge=23.2
0.6
0.4
0.2
0
10
20
30
Photon Energy (keV)
Figure 3. The spectrum of X rays generated with molybdenum and rhodium. On the far left (a) is
the X-ray spectrum generated by a molybdenum target at 30 kVp and filtered with aluminum filter.
The characteristic X rays Ka (17.4 keV), and Kb (19.6 keV) appear as the two peaks in the graph. In
the middle (b) is the X-ray spectrum generated by the same X-ray system in part a, but is filtered
with 30 mm thick molybdenum. The K-absorption edge curve is shown in dashed line. On the far
right (c) is an X-ray spectrum generated by a rhodium target at 30 kVp and filtered with 20 mm thick
rhodium. The general shapes of middle figure and far right figure are similar with differences in the
peak energies of K-characteristic X rays (Ka 20.2 keV, and Kb 22.7 keV), and the K-absorption
edge energy (23.2 keV).
20 keV would be preferentially absorbed than those just

below 20 keV. Similarly, the same can be said for the
rhodium target with rhodium filter as shown in Fig. 3c.
Figure 3b and c are graphically speaking quite similar. It is
noteworthy to point out that a careful study of Fig. 3b, and c
will reveal that X-ray beams generated from rhodium
target X-ray tube must not be filtered with molybdenum!
The intensity of rhodium K-characteristic X rays (Ka, and
Kb) would be in the energy range where the molybdenum K
absorption is high. Taking advantage of the spectral information, in 1991,GE introduced the Senographe DMR unit
equipped with a dual track and dual filter (molybdenum
and rhodium) X-ray tube for mammography applications.
Some of the physical characteristics and the spectral
energy data of molybdenum, rhodium, and tungsten are
summarized in Table 1.
For illustration purposes and descriptions of imaging
components, following this paragraph, the screen-film
mammography (SFM) system manufactured by GE, the
Senogaphe DMR mammography unit, is shown in Fig. 4.
The photograph in Fig. 4 represents the overall external
Table 1. K-Characteristic X Rays and K-Absorption Edge

of Molybdenum, Rhodium, and Tungsten
Atomic Number
Kaa
Kba
K-edgea
a
Energy in keV.
Molybdenum
Rhodium
Tungsten
42
17.4
19.6
20.0
45
20.2
22.7
23.7
74
59.3
69.1
69.5
and mechanical design of a typical dedicated X-ray

mammography unit. It represents the overall design with
respect to the tilting gantry with the X-ray tube housing at
the top and the image receptor at the bottom. And, the
gantry is attached to a column (or stand), which houses the
elevation mechanism of entire gantry.
THE SCREEN-FILM MAMMOGRAPHY

The screen-film mammography employs the same basic
image receptor system as conventional radiographic imaging. While conventional radiography employs a doubleemulsion film sandwiched between two intensifying \
screens yielding a spatial resolution of (up to) 8 lp mm1
for a detailed screen (15), a typical SFM image receptor
consists of a single-emulsion film and a single thin rareearth phosphor intensifying screen yielding a spatial resolution of 20 lp mm1 (16).
In order to optimize the efficiency of the SFM, manufacturers including Agfa, Kodak, Konica, and Fuji, have
produced matching pairs of the intensifying screen and
film specifically for use with mammography. And, to
further improve the sensitometric characteristics of the
screen-film, the processing chemistry, particularly the
developer, have also been carefully prescribed along with
its development conditions. Note that the sensitometric
characteristics of screen-film system refers to the photographic effect or blackening effect of the screen-film
system in response to the X-ray absorption (17). An
example of this matching pair of intensifying screen and
film is depicted in Fig. 5, the emission and absorption
MAMMOGRAPHY
301
Spectral characteristics
20
AD-M film
0.5
10
SLG-8U
AD mammo screens
0
350
400
450
500
550
Wavelength (nm)
600
650
Filter transmittance
Relative sensitivity
1.0
0
700
Figure 5. Spectral characteristics of screen-film mammography

system. The AD Mammo screens are green-emitting and the AD-M
film is orthochromatic. The figure shows the light-emitting spectrum of the AD Mammo Screens, the spectral sensitivity curve of
the AD-M film, and the transmittance spectrum of SLG-8U safety
light filter. (Courtesy of Fujifilm Medical Systems USA, Inc.)
In order to accommodate the varying size of breasts, two

imaging cassette sizes are available with dedicated mammography equipment, they are 18 24 cm (8 10 in.), and
24 30 cm (10 12 in.). The film is loaded on the topside of
the cassette with the emulsion side facing the intensifying
screen. Figure 6 is a schematic drawing showing the crosssection of a typical screen-film mammography cassette (the
single emulsion film, single intensifying screen) with
enlarged views of the film (top, right), and the screen
(bottom, right). The sponge, in the cassette, is employed
to assure good screen-film contact.
THE ANTISCATTER GRID
Figure 4. A typical dedicated mammography unit; GE

Senographe DMR. (Courtesy of GE Healthcare.) The overall
mechanical design of a typical dedicated X-ray mammography
showing the tilting gantry with the X-ray tube housing at the
top and the image receptor at the bottom. The gantry (or the
elongated C arm) is normally attached to a column or a stand in
which the elevation mechanism of entire gantry is housed.
characteristics of intensifying screen and film employed in

SFM. Note that the emission of 550 nm wavelength light
from AD Mammo Screens is matched by the absorption
spectra of the AD-M Film.
The antiscatter grid is placed between the breast holder,

and the cassette slot, refer to Fig. 4. Scattered radiation is
one of the main causes of degrading the image quality in
X-ray imaging. The antiscatter grid is employed to minimize the scattered radiation from reaching the screen-film
system while allowing the primary radiation to pass
through. Although, mammography examinations are typically conducted with X-ray potentials <30 kVp, would still
require a moving (reciprocating) antiscatter grid to cleanup
the scattered radiation. Typically, the antiscatter grid used
in mammography has a grid ratio of 5 : 1, or 4 : 1. While the
exposure time in X-ray mammography imaging is relatively long (1 s), the speed of the moving grid must be
Protective coating
Antihalation layer
Cassette front
Film base
Emulsion
Intensifying screen
Screen base
Sponge
(for film-screen contact)
Cassette back
Film base
Lower emulsion
Upper emulsion
Protective coating
Phosphor layer
(Gd2O2:Tb)
Fine grain
Highly transparent binder
High packing density
Screen base
Figure 6. The cross-sectional view of screenfilm cassette. (Courtesy of Fujifilm Medical

Systems USA, Inc.)
MAMMOGRAPHY
THE COMPUTED RADIOGRAPHY FOR MAMMOGRAPHY

Computed radiography was introduced to radiology in 1981
and the digitization of routine radiographic examinations
had started in earnest. The CR image plate housed in
cassette replaced, directly, the screen-film cassette in routine radiography applications. This direct replacement
design required no mechanical modifications on the existing radiographic equipment. With the introduction of CR, a
whole new set of technology including the optical CR readers, image processing software programs, image display
subsystems, and so on, made the filmless radiology department within an achievable reality (19).
The difference between the routine CR and the CR for
mammography (CR-M) is largely due to the optical
response of the CR phosphor; thus, the radiation dose
required to reduce the image noise, and the spatial resolution capability for mammography applications. In 2001,
Fuji introduced the FCR 5000MA, which was used in the
America College of Radiology Imaging Network (ACRIN);
Digital versus Screen-Film Mammography (DMIST) study
(20,21). While the study had already been concluded, the
official results are in the final preparation stage, and have
not been released yet. The FCR 5000MA differentiated
itself from the previous generation of CR products by
utilizing a 50 mm pixel spot size for the laser and introduced dual side IP reading to the market.
In January 2004, Fuji introduced the Profect CS, or
ClearView-CS in the United States. Clear View CS is
pending FDA approval and not yet commercially available
in the United States. Fuji recently finished the Premarketing Approval (PMA) application to FDA, and the approval
is anticipated sometime during the second half of 2005 (22).
Note, however, that the 100 mm pixel spot size digital
mammography system has been in use for the past few
years in Europe, Australia, and Japan. The mechanism in
which how the IP and the CR reader work together to
produce an image is beyond the scope of this article and
readers are suggested to turn to publications available
(23,24), or corresponding articles in this Encyclopedia.
The obvious advantage of CR-M, and the full field digital
mammography (FFDM) systems, is its wide dynamic range
and its linear response to radiation. In screen-film mammography, over-or underexposure was one of the main
causes of repeat examinations. Such exposure related
repeats can be minimized due to the wide latitude in
exposure acceptable for imaging. Furthermore, the images
are no longer fixed or limited; the subject contrast and
the overall image brightness can be adjusted for image
interpretation by adjusting the display window level, and
Comparison of dynamic range

105
104
Imaging plate (IP)
103
2
102
101
100
Screen-film
Optical density
carefully adjusted to avoid artifact associated with grids.

For the antiscatter grids, the grid line rate must be sufficiently high, or the attenuation material structure must be
so designed to avoid being imaged as grid artifacts on the
mammograms. The honeycomb design antiscatter grid;
Cellular (HTC) Grid utilized in LoRad mammography
systems and marketed by Hologic is well known for its
high transmission of useful primary radiation with
increased absorption of scattered radiation (18).
IP's relative
luminescence
302
102 101 100 101 102

Exposure (mR)
Figure 7. Comparison of dynamic range of CR and screen-film
image receptors. (Courtesy of Fuzifilm Medical Systems, USA,
Inc.) The image plate (IP) has wider linear response to radiation
extends over four decades of dose and eliminates the inherent
limitations of the toe and shoulder portions of the films H&D
Curve. (Courtesy of Fujifilm Medical Systems USA, Inc.)
width. The display window level corresponds to the film

optical density in SFM, and the widow width corresponds
to the contrast. In digital imaging systems, both window
level and width may be adjusted to optimize the image
rendered on the monitor. Figure 7 shows the response differences of these two image receptor systems to the radiation exposure. The exposure range for the CRs Image
Plate is wider than that of a typical screen-film combination by an order of 2 to 3 in magnitude.
THE STEREOTACTIC BREAST BIOPSY MAMMOGRAPHY

Breast biopsy is often required when an area of suspicious
malignancy site is revealed after a mammography is
obtained. In order to accurately extract the specimen, a
stereo mammogram may be obtained so that the biopsy
needle can be accurately inserted to the site where it is
suspected of malignancy; core biopsy (25). Most dedicated
mammography equipment are designed to perform routine
(screen-film, or FFDM) mammography, and with an
attachment to perform stereotactic mammography and
core biopsy in an up right posture; either the patient is
standing or sitting on a chair.
From a pair of stereo images separated by 308 (158
from the centerline), using the triangulation technique, the
three-dimensional (3D) coordinates of the suspicious site
can be calculated within an accuracy of 1 mm (25), see the
schematic diagram of stereotactic imaging geometry in
Fig. 8. It is essential that the patient remain still before
and after the acquisition of the stereo images. The core
biopsy can be localized accurately only if the patient positioning remain the same.
Stereotactic mammography requires that the breast
being examined is compressed (just as in routine mammography), but the patient should remain standing still while
the mammogram is being processed. The processing time
alone takes at least 90 s with typical automatic film processor. Thereafter, the best location of the biopsy needle
entrant site, and the coordinates of the sampling must be
determined. The prolonged time represents substantial
MAMMOGRAPHY
303
+/ 15
Z
Y
X
Compressed breast
Breast holder
Imaging plane/film
Right image
Left image
Target
Figure 8. Schematic drawing of geometrical setup for stereotactic

mammography. From the stereo images, with known geometry
and the XY coordinates, the depth in Z direction is calculated via
triangulation.
discomfort on the part of the patient not to mention the

anxiety of fear for the possibility of being diagnosed and
confirmed as having breast cancer.
DIGITAL SPOT MAMMOGRAPHY
Before the introduction of FFDM, in order to ease the
discomfort of undergoing the biopsy process, a small detector with imaging area of 5 5 cm had been developed by
LoRad, for example, so that the stereo images of the breast
can be displayed immediately after exposures are made.
This is one of the successful biomedical applications that
have resulted from collaborative technology transfer programs between the National Aeronautics and Space
Administration (NASA), the National Cancer Institute
(NCI), and the U. S. Department of Health and Human
Services Office on Womens Health (OWH) (26). Since the
imaging area is only 5 5 cm, it is referred to as Digital
Spot Mammography. The biopsy coordinate is then calculated via the built-in software program with a shorter
overall examination time.
In addition, dedicated breast biopsy systems have been
developed where the patient would be lying on her stomach
(recumbent). The breast under examination is positioned
through a hole in the examination table and aligned with
the X-ray equipment and the biopsy needle gun. The
recumbent core biopsy mammography unit manufactured
by LoRad is depicted in Figs. 9 and 10. LoRad is now one of
the brand names sold under Hologic. A similar recumbent
core biopsy unit, called Mammo Test Biopsy System, is
available from Fisher Imaging.
FULL FIELD DIGITAL MAMMOGRAPHY
The CR-M is a direct replacement of the screen-film cassette. In other words, the IP cassette replaced the screen-
Figure 9. The recumbent stereotactic core biopsy system. A

recumbent design provides a stable patient positioning to
acquire a pair of stereo images for accurate localization of
biopsy site while the patient lays flat on her stomach with
comfort during the entire examination. (Courtesy of Hologic.)
film cassette. Therefore, no major mammography equipment modification would be necessary. In DR on the other
hand, just as in conventional radiography and fluoroscopy
applications, the image receptor system may be built into
the image receptor compartment as an integral part of the
imaging assembly (27,28). Manufacturers had attempted to
physically fit the DR detector assembly, or more accurately;
the Flat Panel (FP) detector assembly, into the space occupied by the SFM cassette. To date, no commercial product of
FP detector assembly has been fitted as a direct physical
replacement of the SFM cassette is available (29). In other
words, due to technical difficulties, there is no FP detector
assembly that is sufficiently compact to fit into the space
designed to accommodate the SFM cassette. The image
acquired with FP detector is transmitted to and processed
by the image processing chain thereafter.
While there are a hand full of FFDM systems either
under development or manufactured for clinical applications, three FFDM units are currently available in the
United States, and are described here (27). The fact that
Figure 10. The zoomed view of the recumbent stereotactic core

biopsy system. In the zoomed view, the breast under the
examination is positioned through the opening in the tabletop.
(Courtesy of Hologic.)
304
MAMMOGRAPHY
Figure 11. Schematic drawing of an

amorphous silicon digital detector.
(Courtesy of GE Healthcare.)
these three systems are designed to cover an imaging field

of at least () 18 24 cm, the smaller cassette size of SFM,
the name; FFDM was assigned. This is to signify the full
size imaging filed coverage as opposed to the small field
detector of LoRads Digital Spot Mammography, or Fischer
Imagings Mammo Test Biopsy System.
There is a trend that mammography equipment is being
transformed to digital format. One reason of the slow
pace of this transformation is the large initial capital
expenditure of FFDM. Both, analog and digital formats
are expected to coexist for many more years to come. The
impact of Fujis CR-M cannot be ignored due to the fact that
it is possible to convert and replace the screen-film cassette
directly. Any of the existing SFM units will be able to join
the digital era.
Currently, there are three FFDM systems that have
received FDA approval and are sold in the United States.
The GE Senogaphe 2000D was approved by FDA on
January 28, 2000 and accredited by ACR in February,
2003. Fischer Senoscan received its FDA approval on
September 25, 2001 with subsequent ACR accreditation
in August, 2003. Hologic/LoRads Selenia received FDA
approval and ACR accreditation, respectively, on October 2,
2002, and in September of 2003 (27).
2000D ushered in the digital mammography era to the

radiology community. Depicted in Fig. 12 is the secondgeneration FFDM unit, Senographe DS, with the X-ray
gantry set to 158 with the stereotactic biopsy needle
assembly attached. Both Senographe systems (2000D
and DS) are equipped with CsI scintillator on an a-Si
photodiodetransistor arrays with pixel size of 100 mm
and an image matrix size of 1920 2304, covering a field
of view 19 23 cm.
FISCHER IMAGINGS SENOSCAN FFDM

On the other hand, Fischer Imaging Corporations answer
to the FFDM is the Senoscan FFDM unit. The unique
feature of Senoscan unit is that it employs a slot mechanism for the X-ray filed collimation that is synchronized to
the detector as the slot and detector assembly sweeps
across the imaging field, see Fig. 13. The detector of
GEs FFDM: SENOGRAPHE 2000D, AND SENOGRAPHE DS

Figure 11 depicts the FP detector developed by General
Electric Medical Systems, and the photo inset (top left)
shows the CesiumIodide (CsI) scintillator crystals. The
incident X rays are absorbed by the CsI crystals, which in
turn, convert the X-ray photon energy to light. The CsI
crystals are deposited in columnar shape so as to minimize
the scatter, and optical diffusion of scintillation lights from
one column to the other. The underlying photodiode
transistor amorphous Silicon (a-Si) arrays is connected
to control data lines, then, converts the light to electrical
signals for further processing (30).
GEs FP detector is an indirect-conversion digital detector system. The detective quantum efficiency (DQE) of this
FP detector system has been shown to be 60% at Zero
Frequency (31,32). Essentially, being the first FFDM system introduced to the commercial market, GEs Senographe
Figure 12. Photograph of GE FFDM unit; Senographe DS with

stereotactic attachment installed. The gantry is tilted to 158 and
prepped for Stereotactic imaging. Note, this is the FFDM version of
similar unit shown in Fig. 1, but with zoomed up view for the
stereotactic attachment. (Courtesy of GE Healthcare.)
MAMMOGRAPHY
X-ray tube
Swing arm pivot point
Filter/collimator
assembly
14.5
.88 cm
in
2977
cmi
Detector
10.5
84 cm
in
21 cmin
1.69
Figure 13. A schematic drawing of the scanning mechanism of

Fischer Senoscan FFDM. (Adapted from Fisher Imaging Sanoscan
users manual with permission. Courtesy of Fischer Imaging.)
305
face without the aid of a scintillator. It is said that a

thickness of 250 mm a-Se photoconductor is adequate to
stop 95% of the X-ray photons in the mammographic
energy range (36). The photon energy is converted to a
pair of electron, which is negatively charged, and a positively charged hole. With bias voltage applied, the signal
is read off by the TFT arrays. Thus, this is a directconversion digital detector system. The detector is an
a-Se photoconductor and TFT arrays with pixel size of
70 mm, covering a field of view 24 29 cm, resulting in
an image matrix size of 3328 4096. A summary of the
detector characteristics for these three FFDM systems is
given in Table 2, (37).
READING MAMMOGRAPHY IMAGES
Senoscan unit consists of a layer of thallium activated CsI

scintillator connected to charge-coupled-device (CCD)
chips via fiber optics (33). Hence, this is also an indirectconversion digital detector system.
The use of slot mechanism also means restricting the
X-ray beams for exposure. It offers an advantage of less
scattered radiation for improved image contrast (34).
Therefore, no antiscatter grid is necessary with Senoscan
unit. The absence of the antiscatter grid compensate for a
less efficient use of the available X rays. In addition, a
tungsten anode X-ray tube (35) is employed with this unit,
since the tungsten anode X-ray tube would produce X rays
more efficiently than a molybdenum or rhodium target.
Senoscan is equipped with CCD chips having pixel size of
(1) 27 mm, covering a field of view 11 15 cm under the high
resolution mode, and (2) pixel size of 54 mm, covering a field
of view 19 29 cm for the normal mode. The image matrix
size of this system is 4096 5625.
Initially, the FFDM images were printed on dry laser

printers and read on high luminance view boxes (ACR
accreditation required of a luminance of > 3000 cd m2)
(38). All three FFDM systems are equipped with workstations where images are soft-copy read. The workstations
are commonly equipped with two 5-megapixel high resolution monitors. With the soft-copy reading, an assortment of
image manipulation are possible including, but not limited
to, basic window width and window level adjustments,
zooming, image reversal, and so on for viewing the details
of the pathology. More importantly, the impact of image
processing in transforming the acquired raw data set to the
image displayed for soft-copy reading should be recognized
(39). For example, as can be seen in Fig. 14, substantial
differences not only in the brightness and contrast of the
image but also in the impression of pathology in the images
can be noticed. While all three images are acceptable and
diagnostic, the two enhanced images are easier to recognize various details.
HOLOGIC/LORADS FFDM; SELENIA
SUMMARY
Selenias image receptor is a 250 mm thick amorphous

Selenium (a-Se) photoconductor. Underneath a layer of
a-Se photoconductor is a thin-film transistor (TFT) arrays
that serves as an active readout mechanism. The TFT
arrays are typically deposited onto a glass substrate, which
provides a physical support for the entire detector components. Selenia uses a 250 mm thick a-Se photoconductor to
capture the X-ray photons impinging on the detector sur-
A brief history of mammography was described as the

introduction to the X-ray mammography. The SFM continues to play its major role in breast cancer detection while
FFDM is gaining its installed base. Upon the anticipated
FDA approval, the CR-M is poised for introduction for
clinical applications in the second-half of 2005. However,
all three modalities employed in breast cancer detection,
namely; the SFM, FFDM, and the CR-M are expected to
Table 2. Summary of Imaging Characteristics of FFDM Units

Manufacturer
Model Name
Scintillator
Detector
Pixel Size, mm
Image Matrix Size
Imaging Size, L W cm
GE
Fisher Imaging
Hologic/LoRad
Senographe 2000D, DS
Senoscan
Selenia
DSM
CR-M
CsI (Tl)a
CsI (Tl)
a-Seb
CsI (Tl)
BaFBr(Eu)c
TFT
CCD
TFT
CCD
Computed
radiography
100
24/48
70
48
50
1920 2304
4096 5625
3328 4096
1025 1024
1770 2370
2364 2964
19 23
22 30
25 29
55
18 24
24 30
Fuji
a
CsI(Tl) Talium Activated Cesium Iodide.

a-Se Amorphous Selenium.
c
BaFBr(Eu) Barium Fluorobromide with Europium.
b
306
MAMMOGRAPHY
Figure 14. Comparison of processed images. The unprocessed

image (a) can be dramatically improve its appearance with
Thickness Equalzation processing (b), or the Premium View
processing (c). The impact of image processing is quite evident.
(Courtesy of GE Healthcare.)
play their roles in contributing to reduction of breast

cancer deaths through the early detection of mammography screening.
Display of FFDM images and its workstations are
important components of the total picture of FFDM
(40,41). However, monitors and workstations associated
with diagnostic radiology imaging including mammography are in a domain of their own and no attempt is made to
include these subjects in this article. Such subject matters
belong to digital image processing and display. Finally,
archiving of the acquired digital images is also a very
important aspect of digitized diagnostic images in the
overall operation of radiology. However, this subject is
better handled in Picture Archiving and Communication
Systems (PACS), and readers are refered to articles under
PACS for additional information.
ACKNOWLEDGMENT
The information and materials presented here had been
provided by numerous numbers of representatives and
scientists from respective manufacturers mentioned in
the main text. The author would like to express his thanks
and sincere appreciation by listing the manufacturers
here, (in alphabetical order) in lieu of listing the individuals who had provided their supports: General Electric Healthcare, Fischer Imaging Corporation, Fujifilm
Medical Systems USA, Inc., and Hologic Inc. (LoRad).
BIBLIOGRAPHY
1. Nystrom L, et al. Breast cancer screening with mammography: Overview of Swedish randomised trial, Lancet
1993; 341:973978.
2. Hendrick RE, et al. Benefit of screening mammography in
women aged 4049: A new meta-analysis of randomized controlled trial. J Nat Cancer Inst Monograph 1997; (22):8792.
3. Mammography-A Users Guide, Washington, DC: National
Council on Radiation Protection and Measurements, 1986.
NCRP Report No. 85.
4. Zermenhof RA. Mammography, Encyclopedia of Medical

Devices and Instrumentation, Vol 3. New York: John Wiley
& Sons; 1988.
5. Jones CH. Methods of breast imaging. Phys Med Biol
1982;27(4)463499.
6. FDA Home page. [Online]. Available at http://www.fda.gov/
cdrh/mammography/frmamcom2.html.
7. ACR Home page. [Online]. Available at http://www.acr.org/
s_acr/sec.asp?CID589&DID14253.
8. Speiser RC, Zanrosso EM, Jeromin LS. Dose comparisons
for mammographic systems. Med Phys 1986;13(5):667
673.
9. Boag JW. Xeroradiography. Phys Med Biol 1973;18:3.
10. Wolbarst AB. Dependence of attenuation on atomic number
and photon energy. Physics of Radiology. Appleton & Lange;
1992. Chapt. 14.
11. Hoeffken W. Soft tissues-mammography. The invisible light
applied, the advancing diagnostic evaluation of the skeleton
and thorax following Roentgens discovery. In: Rosenbusch G,
Oudekerk M, Ammann E, editors. Radiology in Medical
Diagnostics, Evolution of X-ray Applications 18951995.
Oxford: Blackwell Science Ltd.; 1995. Chapt. 2.
12. Curry III TS, Dowdey JE, Murry Jr. RC. Christensens
Physics of Diagnostic Radiology, 4th ed., Philadelphia: Lea
& Febiger; 1990.
13. Fewell TR, Shuping RE. Handbook of Mammographic X-ray
Spectra. Rockville, MD: HEW Publication (FDA) 798071;
1978.
14. Boone JM, Fewell TR, Jennings RJ. Molybdenum, rhodium,
and tungsten anode spectral modeling using interpolating
polynomials with application to mammography. Med Phys
Dec. 1997:24(12).
15. Sprawls Jr. P. Chapter 18: Blur, Resolution, and Visibility of
Detail. Physical Principles of Medical Imaging. Aspen Publishers, Inc.; 1987.
16. Haus AG. Technologic improvements in screen-film mammography. Radiology 1990;174:628637.
17. Chapter V. Film Response. The Fundamentals of Radiography, 4th ed., Rochester, N.Y.: Eastman Kodak Co.; 1980.
18. Rezentes PS, de Almeida A, Barnes GT. Mammography grid
performance. Radiology 1999;210:227.
19. Sonoda M, Takano M, Miyahara J, Kato H. Computed radiography utilizing laser stimulated luminescence. Radiology
1983;148:833838.
20. Pisano ED, et al. American College of Radiology Imaging
Network Digital Mammographic Imaging Screening Trial:
Objectives and Methodology. Radiology published online.
June 16, 2005.
21. ACRIN Home Page. [Online]. Available at http://www.acrin.
org.
22. Communication with National Program Manager, Womens
Healthcare Imaging Systems, Fuji Photo Film Ltd.
23. FCR (Fuji Computed Radiography) General Description of
Image Processing, Fuji Photo Film Co., Ltd.
24. Tateno Y, Iinuma T, Takano M, editors. Computed Radiography. Tokyo: Springer-Verlag; 1987.
25. Carr JJ, et al. Stereotactic localization of breast lesions: How
it works and methods to improve accuracy. RadioGraphics
2001;21:463.
26. Winfield DL. Aerospace technology transfer to breast cancer
imaging. Acta Astronau 1997;41(410):515523.
27. Pisano ED, Yaffe MJ. Digital Mammography. Radiology
2005;234:353362.
28. Pisano ED. Current status of full-field digital mammography.
Radiology 2000;214:26.
29. Personal communications with various X-ray industry marketing departments.

30. Vedantham S, et al. Full Breast digital mammography with
anamorphous silicon-based flat panel detector: Physical
characteristics of a clinical prototype. Med Phys 2000;
27(3):558567.
31. Albagli D, et al. Performance of advanced a-Si/CsI-based flat
panel X-ray detectors for mammography. In: Yaffe MJ, Antonuk LE, editors. Proceedings of SPIE. Medical Imaging 2003:
Physics of Medical Imaging, Vol. 5030, June 2003. p 553563.
32. Shaw J, Albagil D, Wei C-Y. Enhanced a-Si/CsI-based flat
panel X-ray detector for mammography. In: Yaffe MJ, Flynn
MJ, editors. Proceeding of SPIE. Medical Imaging 2004:
Physics of Medical Imaging. Vol. 5368, May 2004. p 370378.
33. Tesic MM, Picaro MF, Munier B. Full field digital mammography scanner. Eur J Rad 1997;31:217.
34. Boone JM, et al. Grid and slot scan scatter reduction in
mammography: Comparison by using Monte Carlo techniques. Radiology 2002;222:519527.
35. Operator Manual, SenoScan: Full Field Digital Mammography System, Fischer Imaging Corporation, Denver, CO, Dec.
2002.
36. Smith AP. Fundamentals of digital mammography: Physics,
technology and practical considerations. Radiol Manager
2003, Sep.Oct.; 25(5):1824, 2631.
37. Mahesh M. AAPM/RSNA physics tutorial for residents,
Digital mammography: an overview. Radiographics
2004;24: 17471760.
38. The ACR Mammography Quality Control Manual. Preston,
VA: ACR, 1999.
39. Pisano ED, Yaffe MJ. Digital mammography. Radiology
2005;234:353362.
40. Hemminger BM, et al. Evaluation of the effect of display
luminance on the feature detection of simulated masses in
mammograms. Proc SPIE 1997;3036:12.
41. Pisano ED, et al. Radiologists preferences for digital mammographic display. Radiology 2000;216:820830.
Further Reading
Mahesh M, AAPM/RSNA physics tutorial for residents, Digital
mammography: An overview. Radiographics 2004;24:1747
1760.
Smith AP, Hall PA, Marcello DM. Emerging technologies in breast
cancer detection. Radiology Manager 2004, July.Aug.; 26(4):
1624.
Shramchencko N, Blin P, Mathey C, Klausz R. Optimized exposure
control in digital mammography. In: Yaffe MJ, Flynn MJ,
editors. Proceedings of SPIE. Medical Imaging 2004: Physics
of Medical Imaging. May 2004; Vol. 5368, p 445456.
Curry III TS, Dowdey JE, Murry Jr. RC. Christensens Physics of
Diagnostic Radiology. 4th ed., Philadelphia: Lea & Febiger;
1990. (This is a textbook used in various radiology residency
programs across the United States.)
Online References
FDA Home page. [Online]. Available at http://www.fda.gov/cdrh/
mammography/frmamcom2.html.
ACR Home page. [Online]. Available at http://www.acr.org/.
ACRIN Home Page. [Online]. Available at http://www.acrin.org.
GE Healthcare Home Page. [Online]. Available at http://www.
gehealthcare.com/usen/whc/whcindex.html.
Fischer Imaging Home Page. [Online]. Available at http://www.
fischerimaging.com/default/default.asp.
Fujifilm Medical Systems USA Home Page. [Online]. Available at
http://www.fujimed.com.
Hologic Home Page. [Online]. Available at http://www.hologic.
com/.
307
MATERIALS, BIOCOMPATIBILITY OF. See

BIOCOMPATIBILITY
OF MATERIALS.
MATERIALS, PHANTOM, IN RADIOLOGY.

PHANTOM
See
MATERIALS IN RADIOLOGY.
MATERIALS, POLYMERIC. See POLYMERIC MATERIALS.

MATERIALS, POROUS. See POROUS MATERIALS FOR
BIOLOGICAL APPLICATIONS.

ARIE HASMAN
Maastricht
The Netherlands
INTRODUCTION
The amount of knowledge is increasing rapidly in many
disciplines. Medicine is not an exception. Because of scientific research new knowledge comes available at such a
pace, that physicians should read 19 articles a day, every
day of the week, to keep up to date. Since that is not
possible the results of scientific research often are applied
clinically only years later. Computers can support physicians in finding relevant recent information. In the next
section, the reasons for using computers in medical education are presented. Then the roles computers can play in
medical education are reviewed. Since health professionals increasingly use computer systems for their work,
they need to know the benefits and limitations of these
systems. The discipline of medical informatics is responsible for developing these systems and therefore is discussed. The next sections discuss the use of Internet and
electronic patient records. Also, it is discussed why knowledge of information systems is important for health professionals.
THE REASONS FOR USING COMPUTERS

IN MEDICAL EDUCATION
The goal of academic medical education is to educate
students to become physicians. During the study knowledge, skills and attitudes have to be mastered. Students
are taught academic skills like critical reading, they are
acquainted with the principles of research methods, and
they should know the scientific background of the basic
disciplines (like anatomy, molecular cell biology and genetics, endocrinology and metabolism, immunology and
inflammation, growth, differentiation and aging) as far
as they are related to the study of abnormalities and to
diagnosis and therapy. Because of the explosive growth of
biomedical knowledge, it is not possible anymore to teach
all the currently available knowledge. This does not have to
be a problem since part of the knowledge presented to
medical students during their formal education may be
more or less obsolete by the time they are in their main
professional practice. Moreover, it is difficult to teach
medicine for the coming era, since most of the futures
308
technology is probably nonexistent today. Students therefore must be taught how to become life-long learners.
Computers can support this process.
Computers play an increasing role in the practice of
medicine too. No doctorwhether a general practitioner or
a specialist in advanced or social carewill be able to
escape the confrontation with some form of information
processing. The work of health professionals is dominated
by information collection, storage, retrieval, and reasoning. Health professionals both use individual patient data
and general medical or nursing knowledge. The amount of
medical and nursing knowledge increases so quickly that
health professionals cannot stay fully up to date. Tools are
therefore needed to acquire relevant knowledge at the time
it is needed.
Computer systems are installed in many hospital
departments and physician offices. Hospital information
systems support, for example, financial, administrative,
and management functions. Clinical departmental systems are used to collect, store, process, retrieve, and communicate patient information. Clinical support systems
are used in function laboratories [for electroencephalogram
(EEG), electrocardiogram (ECG), electromyogram (EMG),
and spirometry analysis], for imaging [magnetic resonance
imaging (MRI), computerized tomography (CT), nuclear
medicine, ultrasound], and in clinical laboratories (analysis of electrolytes, etc.). The results of clinical support
systems are increasingly stored in so-called electronic
patient records, together with the medical history, results
of physical examination, and progress notes. Electronic
patient records gradually replace the paper patient record.
Apart from the fact that electronic paper records are better
readable they also support functions (like decision support)
paper records cannot provide. Students must learn the
benefits and limitations of these kinds of systems.
Decision support systems are used to support clinicians
during the diagnostic or therapeutic process and for preventive purposes to prevent either errors of omission
(when, eg, the physician does not order a mammography
when indicated) or commission (when, eg, the physician
prescribes the wrong drug).
Clinical patient data are increasingly stored in the
above mentioned electronic patient records (EPRs), from
which they can be later retrieved by physicians or nurses
who are in need of the data. Also, information systems can
retrieve relevant data from the electronic patient record
when a suitable interface between system and EPR is
available. When a standard vocabulary is used for representing data values in the EPR, decision support systems
can interpret these data and remind, alert, or critique the
physician or provide access to relevant knowledge, based
on patient data available in the EPR. Health professionals
should have insight in and knowledge of the principles,
concepts, and methods underlying electronic patient
records.
Also, patients become active players in the field and
increasingly demand access to the EPR.
Patient management increasingly has become the combined task of a group of healthcare workers. Therefore the
memory-aid role of the patient record more and more
changes into a communication role. Paper records have
several limitations in this respect. In addition, since the

appearance of the report To err is human of the IOM (1) it
is apparent that due to miscommunication (among which
are problems with reading handwritten information, with
incomplete information, etc.) medical errors are made that
even may lead to the death of patients. Electronic patient
records and order entry systems can reduce the number of
errors because they are not only more readable, but
because they can also be interfaced with decision support
systems when standardized terminology is used.
Decision support can be passive in the sense that the
decision support system contains information that has to
be searched by the physician. In this case, the healthcare
professional takes the initiative to search for information,
for example, via PubMed or the Cochrane Library. Decision
support can also be active. In this case, the decision support
system volunteers advice based on information it can
retrieve from the EPR. Decision support systems can either
proactively suggest the next step in a diagnostic or treatment process or reactively remind the healthcare professional that a (preventive) procedure was not performed or a
step in the protocol was not carried out.
ROLES FOR COMPUTERS IN MEDICAL EDUCATION

What roles can computers play in medical education? In
the first place, information systems can be used to manage
the learning process. Students can get access to the curriculum via so-called learning environments (e.g., Blackboard or WebCT), can get overviews of their marks, can
access computer aided instruction programs, can access
PowerPoint presentations of their teachers, and so on.
Computers provide access to the internet so that they
can search for content knowledge.
Computer-aided instruction can be used to teach students certain subjects. For example, computers are used to
simulate regulatory mechanisms occurring in the human
body. With a simulation program, students can treat
patients without risking their patient lives. The simulation is often based on models that present (patho)physiologic processes in the form of mathematical equations.
When the models become increasingly accurate they can
even be used in patient care. Also, patient management
problems can be simulated. In this case, usually no mathematics is involved, but the patients signs and symptoms as
they develop as a function of time are expressed as
text.
There are simulation tools that allow users to evaluate,
plan, or redesign hospital departments, or parts of other
healthcare systems. Physicians will be confronted with the
results of simulations. The model that is used in the
simulation has to be checked for validity. Some tools present the modeled processes visually so that physicians or
nurses can easily determine the correctness of the model.
An example is the modeling of a phlebography service. On
the screen, the cubicles and other rooms are displayed and
the movements of patients, physicians, and nurses can be
followed. In this way, the users can judge whether the
model represents the situation of a phlebography service in
an adequate way.
Note that computers should not be considered as surrogate teachers controlling students learning. Computers
should enrich the learning environment by expanding
the students control over their self-learning and by providing a better learning environment as a supplement to
traditional methods of learning. The effectiveness of CAI
has always been a subject of controversy. Studies have
claimed both that CAI is superior and that CAI is inferior to
traditional methods. The majority of the publications, however, support the notion that CAI is as effective as traditional educational methods (2).
Although decision making is the pre-eminent function of
a physician, hardly anywhere is the student confronted
with a systematic exposition of procedures of good decision
making. The use of computers can facilitate the teaching of
these procedures. Computer support offers new possibilities to teach problem solving techniques and analytical
methods that are presently learned by the student through
practice and the observation of mentors.
MEDICAL INFORMATICS
Education concerning the advantages and limitations of
the use of computers for supporting the work of health
professionals is the responsibility of medical informatics
departments. Medical informatics can also be instrumental
in developing computer-aided instruction programmes and
simulation packages. Medical informatics is the discipline
that deals with the systematic processing of data, information, and knowledge in the domains of medicine and healthcare. The objects of study are the computational and
informational aspects of processes and structures in medicine and healthcare (3). Medical informatics is a very
broad discipline covering subjects like applied technology
(bioinformatics, pattern recognition, algorithms, human
interfaces, etc.) and services and products (quality management, knowledge-based systems, electronic patient
records, operationsresource management, etc.). Also
human and organizational factors (managing change, legal
issues, needs assessment, etc.) should be taken into
account.
Informatics can be either a systems-oriented discipline
in which computer systems, operating systems and programming languages are the object of study or a methodsoriented discipline in which the methods are studied that
can be used to create algorithms that solve problems in
some application domain. In the case of the methodsoriented approach, a problem is studied and formalized
solutions are determined. Medical informatics is an example of this approach: it studies the processing of data,
information, and knowledge in medicine and healthcare.
Medical informatics focuses on the computational and
informational aspects of (patho)physiological processes
occurring in the patient, cognitive processes going on in
the brain of physicians, and organizational processes that
control healthcare systems.
The resulting knowledge can be used to design information systems that can support healthcare professionals. It
is clear that healthcare professionals need to have some
medical informatics knowledge in order to optimally use
309
information systems. Medical informatics education

should therefore be part of the medical curriculum.
Various groups of professionals with quite different
backgrounds can be identified who carry out medical informatics tasks ranging from the use of IT to developing
information systems. Users of information systems naturally need less medical informatics knowledge than health
informatics experts who develop health information systems or support other healthcare workers in designing
terminology servers, coding systems, and so on.
There exists a wide range of job opportunities in the field
of medical informatics. These jobs require various medical
informatics capabilities. In addition to medical informatics,
students (and graduate students) with other backgrounds
may prefer a job in the field of medical informatics. In order
to obtain the relevant capabilities these students have to
learn additional subjects depending on their previous education and the type of specialization they want to achieve.
These students can be graduates from healthcare related
programs or from informaticscomputer science programs.
Graduates from healthcare related programs possess the
relevant medical knowledge, but need to increase their
medical informatics knowledge. Graduates with an informatics or computer science background must learn how the
healthcare system is organized and how healthcare professionals are working in order to develop systems that are
appreciated by healthcare professionals. Medical informatics is therefore taught in different ways (4) depending
on the type of students and the type and extent of specialization that they want to achieve.
USE OF THE INTERNET

Much knowledge can be found on the internet. Browsers
allow health professionals and patients to access sites
containing (references to) medical knowledge. PubMed is
an example. It contains references to the medical literature. The web contains a lot of information of which the
quality is not always guaranteed. Especially in the medical
arena, this is a big disadvantage. The internet has become
one of the most widely used communication media. With
the availability of Web server software, anyone can set up a
Web site and publish any kind of data that is then accessible to all. The problem is therefore no longer finding
information, but assessing the credibility of the publisher
as well as the relevance and accuracy of a document
retrieved from the net. The Health On the Net Code of
Conduct (HONcode) has been issued in response to concerns regarding the quality of medical and health information (5). The HONcode sets a universally recognized
standard for responsible self-regulation. It defines a set
of voluntary rules to make sure that a reader always knows
the source and the purpose of the information they are
reading. These rules stipulate, for example, that any medical or health advice provided and hosted on a site will only
be given by medically trained and qualified professionals
unless a clear statement is made that a piece of advice is
offered from a nonmedically qualified individual or organization. Another guideline states that support for the Web
site should be clearly identified, including the identities of
310
commercial and noncommercial organizations that have

contributed funding, services or material for the site. Students searching for information should be introduced to
these guidelines.
Searching can be carried out by entering keywords.
These keywords can be connected by Boolean operators
like AND, OR, and NOT. A user can for example enter:
Diuretics and Hypertension to search for documents that
discuss the use of diuretics in hypertension. The NLM
(National Library of Medicine) uses the Medical Subject
Headings (MeSH) vocabulary for indexing most of their
databases. Students should be taught how to efficiently
search in bibliographic databases using, for example, the
MeSH vocabularies.
We speak of e-learning when content is accessible via
Web browsers. Some characteristics of e-learning follow:
Internet is the distribution channel. Access to the content
is possible 24 h/7 days a week. It is learner-centered. The
student determines the learning environment, the speed of
learning, the subjects to consider, the learning method. A
mix of learning methods can be used (blended learning): for
example, virtual classroom, simulations, cooperation, communities, and live learning.
Virtual learning environments aim to support learning
and teaching activities across the internet. Blackboard and
WebCT are examples of such environments. These environments offer many possibilities. New or modified educational modules can be announced or teachers can give
feedback regarding the way a module is progressing. Also
general information about a module can be provided. Staff
information can be presented with photo, email address,
and so on. Assignments can be posted, and so on. The
virtual classroom allows students to communicate online,
whereas discussion boards allow asynchronous communication. Also, links to other websites can be provided. The
internet is a source of information for patients. They can
retrieve diagnostic and therapeutic information from the
internet. Increasingly, patients present this information to
their care providers. Health professionals must know how
to cope with this new situation and must be able to assess
the quality of the information.
KNOWLEDGE OF INFORMATION SYSTEMS

Information systems are increasingly used in healthcare.
They not only support administrative and financial, but
also clinical and logistic processes. Since healthcare workers have to use information systems they should know the
possibilities, but also the limitations, of information systems. Since in information systems, for example, data can
be easily retrieved, the quality of entered data determines
the quality of the results: garbage in, garbage out. In
addition, they should have the skills to work with information systems. Information systems relevant for healthcare
professionals include hospital information systems,
departmental systems, electronic patient record systems,
order entry and result reporting systems, and so on. But
healthcare workers should also be proficient in the use of
productivity tools like word processing systems, bibliographic search systems, and so on.
Logistics is becoming more important these days. Hospitals have to work not only effectively, but also more
efficiently, thereby taking the preferences of patients into
account. Planning systems can reduce the time that ambulatory patients have to spend in the hospital for undergoing
tests, but also the length of stay of hospitalized patients can
be reduced by planning both the patients and the needed
capacity (6).
It is important for healthcare workers to know what
support they can expect from information systems and to
know which conditions have to be satisfied in order that
information systems can really be of help. Optimal use of
information systems therefore does not only depend on
acquired skills, but also on the insight in and knowledge
of the principles, concepts, and methods behind information systems. This is true for all types of healthcare professionals. When hospitals or physicians consider the
purchase of information systems they must be able to
specify their requirements so that they will not be confronted with systems that do not perform as expected.
ELECTRONIC PATIENT RECORDS

Physicians store information about their patients in
patient records. The patient record frequently is a paper
record in which the physician writes his notes. Paper
records have several advantages because they are easy
to use, easy to carry, and so on. But there are also limitations: they may be difficult to read and are totally passive:
the physician records the information with little support
(e.g., headings in a form) and therefore the recordings are
often incomplete. Not only are the data incomplete, they
also contain errors, for example, due to transcription or
because the patient gave erroneous information. The readers of patient records can interpret the data in the patient
record incorrectly. The fact that data are recorded in
chronological order makes the retrieval of facts sometimes
difficult: such data are not recorded on standard positions
in the record. A study showed that because of the constrained organization of paper records, physicians could
not find 10% of the data, although these data were present
in the paper record (7). The patient data are usually stored
in more than one type of record, because each department
uses its own records, each with a different lay-out. Paper
records are not always available at the time they are
needed. Paper records are passive: they will not warn
the physician if he overlooks some results. If the results
of a lab request are unexpectedly not yet available, the
paper record will not indicate that. Despite these drawbacks physicians are usually very positive about the use of
the paper record, because they do not recognize their
shortcomings.
Electronic patient records have some advantages over
paper records. The legibility of electronic patient records is
per definition good. Data can be presented according to
different views (time-, source-, and problem-oriented),
making the data easier to access. Electronic patient
records, when interfaced with decision support systems,
can provide reminders when a physician forgets something.
MEDICAL ENGINEERING SOCIETIES AND ORGANIZATIONS
The use of computers in medical education is diverse.

They can be used for managing the learning process, for
distributing content, for assessing the students knowledge
or skills, and so on. In this case, they can be regarded as
educational tools. As is clear from the above information
systems are extensively used in the practice of health
professionals. Students should be taught the benefits,
but also the limitations of the use of information systems.
Finally the information systems have to be developed. To
be able to do so students need additional education in
medical informatics.
BIBLIOGRAPHY
1. Institute of Medicine. To err is human: Building a safer
health system. The National Academies Press; 2000.
2. Qayumi AK, et al. Comparison of computer-assisted instruction (CAI) versus traditional textbook methods for training in
abdominal examination (Japanese experience). Med Ed
2004;38:10801088.
3. Hasman A, Haux R, Albert A. A systematic view on
medical informatics. Comp Meth Prog Biomed 1996;51:
131139.
4. Haux R, Grant A, Hasman A, Hovenga E, Knaup P. Recommendations of the International Medical Informatics Association (IMIA) on education in health and medical
informatics. Methods Inf Med 2001;40:7882.
5. http://www.hon.ch (last visited 21 December 2004).
6. van Merode GG, Groothuis S, Hasman A. Entreprise
resource planning for hospitals. Int J Med Inform 2004;73:
493501.
7. Fries JF. Alternatives in medical record formats. Med Care
1974;12:871881.
MEDICAL ENGINEERING SOCIETIES AND

ORGANIZATIONS
ARTHUR T JOHNSON
University of Maryland
College Park, Maryland
PATRICIA I HORNER
Landover, Maryland
INTRODUCTION
Modern technology has transformed the practice of medicine. We can now see where we could not before, conduct
surgery with minimal trauma, intervene at the genetic
level, replace whole natural organs with functional artificial ones, make rapid diagnoses, and peer into the workings
of the brain. More patients are surviving, and those who do
are living better. Much of the credit for these advances goes
to the engineers, physicians, and physiologists who
together decided what needed to be done, the science
required to support it, and how it could be made practical.
Medical engineers are now very much involved in the
process of developing medical advances. They bring to
medicine the abilities of conceptualization, computation,
311
and commercialization. They use varied tools such as

biophysics, applied mathematics, physiological modeling,
bioinstrumentation and control, imaging, and biomechanics to accomplish their advances.
The result is that there are nearly as many subspecialties of medical engineering as there are medical specialties. Tissue engineers, for instance, grow bioartificial
tissues and organs as replacements; metabolic engineers
find means to adjust cellular metabolic pathways to produce greater quantities of biochemicals and hormones;
and rehabilitation engineers design new prostheses or
modify existing units to reestablish adequate function
in patients who have lost ability usually as the result of
trauma. There are medical engineers working with biosensors, bioprocess optimization, multiple imaging modes,
pancreatic function, vascular replacement, and drug
delivery. Biomaterials engineers have produced materials
that can function in different regional corporal environments. Indeed, there is no part of the human body that has
not been studied by medical engineers to improve or
replace lost function.
As the body of medical knowledge has increased overall
and has been repeatedly split more and more finely into
specialties, there has been a concomitant proliferation of
organizations to communicate, share, and advocate action
related to their particular specialties. Some of these would
be recognized as chiefly engineering organizations with
application interests in medicine; some are medical societies with significant engineering contributions. There is
almost no significant human disease, physiological system,
organ, or function without a group or organization representing associated interests. There is even a group interested in developing synthetic biological forms that,
although it is too premature to link with medicine, may
someday have a profound effect on medicine. All of these
groups can be found by searching the Internet, and any
attempt to enumerate them here would be outdated very
quickly.
DEFINITIONS
Progress in biological science and engineering has not been
made with a clear distinction between medical and nonmedical applications. Advances in human medicine often
find applications as well in veterinary medicine. Genetic
coding techniques have been applied equally to humans
and fruit flies. Prospective biomaterials are modeled on
computer without regard for the ultimate specific application, and they are tested in animals, plants, or fungi before
approval for human use. Progress toward better nutrition
through science, and toward purer environments through
improved pollutant detection monitoring, have resulted in
better human health for most humans living in the developed world. Biology is biology, whether applied to human
health care or not, so a convergence of basic knowledge and
methods between medical and biological engineers is
expected to continue.
Several relevant definitions attempt to distinguish
among various fields where engineers have and will continue to contribute.
312
The U.S. National Institutes for Health (NIH) has the

following definition of bioengineering:
Bioengineering integrates physical, chemical, mathematical,
and computational sciences and engineering principles to
study biology, medicine, behavior, and health. It advances
fundamental concepts; creates knowledge from the molecular
to the organ systems levels; and develops innovative
biologics, materials, processes, implants, devices, and informatics approaches for the prevention, diagnosis, and treatment of disease, for patient rehabilitation, and for improving
health.
The U.S. National Science Foundation program in Biochemical Engineering and Biotechnology (BEB) describes
its program in the following way:
Advances the knowledge base of basic engineering and scientific principles of bioprocessing at both the molecular level
(biomolecular engineering) and the manufacturing scale (bioprocess engineering). Many proposals supported by BEB programs are involved with the development of enabling
technologies for production of a wide range of biotechnology
products and services by making use of enzymes, mammalian,
microbial, plant, and/or insect cells to produce useful biochemicals, pharmaceuticals, cells, cellular components, or cell composites (tissues).
The Whitaker Foundation definition of biomedical engineering is as follows:

Biomedical engineering is a discipline that advances knowledge in engineering, biology, and medicine, and improves human health through cross-disciplinary activities that integrate
the engineering sciences with the biomedical sciences and
clinical practice. It includes: 1) The acquisition of new knowledge and understanding of living systems through the innovative and substantive application of experimental and analytical
techniques based on the engineering sciences, and 2) The
development of new devices, algorithms, processes, and systems that advances biology and medicine and improve medical
practice and health care delivery.
And, finally, the Institute of Biological Engineering

(IBE) defines biological engineering as follows:
Biological engineering is the biology-based engineering discipline that integrates life sciences with engineering in the
advancement and application of fundamental concepts of biological systems from molecular to ecosystem levels. The emerging discipline of biological engineering lies at the interfaces of
biological sciences, engineering sciences, mathematics and computational sciences. It applies biological systems to enhance the
quality and diversity of life.
HISTORICAL DEVELOPMENTS
In 1948, in New York City, a group of engineers from the
Instrument Society of America (ISA) and the American
Institute of Electrical Engineers (AIEE), with professional
interests in the areas of X-ray and radiation apparatus
used in medicine, held the First Annual Conference on
Medical Electronics. Soon thereafter the Institute of Radio
Engineers (IRE), joined with the ISA and AIEE, and the
series of annual meetings continued. Subsequent years

witnessed a remarkable growth of interest in biomedical
engineering and participation by other technical associations. By 1968 the original core group evolved into the Joint
Committee on Engineering in Medicine and Biology
(JCEMB), with five adherent national society members:
the Instrument Society of America (ISA), the Institute of
Electrical and Electronics Engineers, Inc. (IEEE), the
American Society of Mechanical Engineers (ASME), the
American Institute of Chemical Engineers (AIChE),
and the Association for the Advancement of Medical
Instrumentation (AAMI), who jointly conducted the
Annual Conference on Engineering in Medicine and Biology (ACEMB).
Professional groups responded vigorously to the
demands of the times. Attendance at the annual conference
by natural scientists and medical practitioners grew to
approximately 40% of the total; medical associations
requested formal participation with their technical counterparts on the JCEMB. New interdisciplinary organizations were formed. New intrasociety and intersociety
groups, committees, and councils became active; meetings
filled the calendar; and publications overflowed the
shelves.
In 1968, a document was prepared that read as follows:
WHEREAS:
1. Common interdisciplinary purposes cannot be well
served by individual groups working independently
from each other;
2. Certain associations have developed in attempts to
meet the need;
3. Conferences and publications have proliferated in
attempts to meet the needs;
4. At present, no mutually satisfactory mechanism
exists for the coordination of the relevant groups
and functions;
5. There does exist an annual meeting and proceedings
publication sponsored by a limited number of societies through the Joint Committee on Engineering in
Medicine and Biology (JCEMB);
6. The JCEMB is formally structured with a constitution, plural societal representati9on, and an established pattern of operation. This structure and
pattern of operation, however, are not deemed adequate to fulfill present and future needs. To the
best of our knowledge, there exists no other single
organization that seems capable of fulfilling these
needs.
THEREFORE, it is appropriate that a new organization
be established.
On July 21, 1969, at the 22nd ACEMB in Chicago,
Illinois, representatives of 14 national engineering, scientific, and medical associations founded the Alliance for
Engineering in Medicine and Biology (AEMB). It was
incorporated on December 24, 1969, in Washington, D.C.
Lester Goodman, Ph.D., served as Founder President in
1970197l; Arthur C. Beall, MD(1972); Alan R. Kahn,

MD(1973); Harry S. Lipscomb, MD(1974); Anthony Sances,
Jr., Ph.D. (1975); Charles Weller, MD(19761977); Edward
J. Hinman, MD MPH(19781979); Paul W. Mayer,
MD(19801982); Francis M. Long, Ph.D., (19831984);
Arthur T. Johnson, PE, Ph.D. (19851988); and Alfred
R. Potvin, PE Ph.D. served as the final President in
19891990.
The Alliance operations were determined by an Administrative Council composed of delegates from each of its
affiliates. Later the Alliance was to consist of more than 20
such organizations:
Aerospace Medical Association (ASMA)
American Academy of Orthopaedic Surgeons (AAOS)
American Association of Physicists in Medicine
(AAPM)
American College of Chest Physicians (ACCP)
American College of Physicians (ACP)
American College of Radiology (ACR)
American College of Surgeons (ACP)
American Institute of Aeronautics and Astronautics
(AIAA)
American Institute of Biological Sciences (AIBS)
American Institute of Chemical Engineers (AIChE)
American Institute of Ultrasound in Medicine
(AIUM)
American Society for Artificial Internal Organs
(ASAIO)
American Society for Engineering Education (ASEE)
American Society for Hospital Engineers of the
American Hospital Association (ASHE)
American Society for Testing and Materials
(ASTM)
American Society of Agricultural Engineers (ASAE)
American Society of Civil Engineers (ASCE)
American Society of Heating, Refrigerating and Air
Conditioning Engineers (ASHRAE)
American Society of Internal Medicine (ASIM)
American Society of Mechanical Engineers (ASME)
Association for the Advancement of Medical Instrumentation (AAMI)
Biomedical Engineering Society (BMES)
Institute of Electrical and Electronics Engineers (IEEE)
Instrument Society of America (ISA)
National Association of Bioengineers (NAB)
Neuroelectric Society (NES)
RESNARehabilitation Engineering & Assistive Technology Society of North America
Society for Advanced Medical Systems, now American
Medical Informatics Association (AMIA)
Society for Experimental Stress Analysis (SESA)
SPIEInternational Society for Optical Engineering
Alpha Eta Mu BetaNational Biomedical Engineering
Student
313
Honor Society, established under the auspices of A

EMB.
The Alliance headquarters office opened on November 1,
1973. John H. Busser served as the first Executive Director. Patricia I. Horner served as Assistant Director, as
Administrative Director, and succeeded Busser as the
Executive Director. Among its goals, is the following
excerpted in part from its constitution, bylaws, and
recorded minutes:
to promote cooperation among associations that have an active
interest in the interaction of Engineering and the physical
sciences with medicine and the biological sciences in enhancement of biomedical knowledge and health care.
to establish an environment and mechanisms whereby people from relevant various disciplines can be motivated and
stimulated to work together
to respond to the needs of its member societies, as expressed by
their delegates, rather than to seek authoritative preeminence
in its domain of interest. . .
to support and enhance the professional activities of its
membership. . .
The 23rd ACEMB in Washington, D.C., in 1970, was the

first held under the aegis of the Alliance. From 1979 to
1984, the IEEE Engineering in Medicine and Biology
Society (EMBS) held their conferences immediately preceding the ACEMB. The Society for Advanced Medical
Systems, later to become AMIA, and the Biomedical Engineering Society also held their meetings for several years
in conjunction with the ACEMB.
The accomplishments of the Alliance far outstripped the
expectations of its founders. The Alliance more than fulfilled responsibilities for the annual conference inherited
from the predecessor JCEMB, but the Alliance made
important contributions through a variety of studies and
publications ranging from a 5-year ultrasound research
and development agendum to a guideline for technology
procurement in health-care institutions:
First International Biomedical Engineering Workshop Series held in Dubrovnik, Yugoslavia, under
the sponsorship of the National Science Foundation.
This project was in cooperation with AIBS and the
International Institute of Biomedical Engineering in
Paris. Five workshops were held, and planning handbooks were completed.
Assessment of selected medical instrumentation;
Tasks 14, ultrasonic diagnostic imaging; Task 5,
radiologic and radionuclide imaging technology.
Summary guidelines and courses on technology procurement; practices and procedures for improving
productivity in research and health-care institutions.
Information exchange and problem assessments in
medical ultrasound, including preparation and distribution of a directory of federal activities, conducted
instrumentation conferences, delineated training
needs, assessed technology transfer potential, and
prepared guidelines for the establishment of clinical
ultrasound facilities.
314
Joint U.S.Egypt international technology transfer

project in medical diagnostic ultrasound, including
international workshops and the design and support
of a focus laboratory for ultrasonic diagnosis at Cairo
University Medical School.
Short courses for continuing education at the
annual conference on engineering in medicine and
biology.
International directory of biomedical engineers.
Before long, the proliferation of medical engineers, and
competing interests among societies, led to a fragmentation of the field. It became clear that the Alliance no longer
represented positions of the entire field. No organized
group could speak for the entire profession, and the spirit
of unity that had led to the development of AEMB no longer
existed. It was time for a new beginning.
AMERICAN INSTITUTE FOR MEDICAL
AND BIOLOGICAL ENGINEERING
In 1988, the National Science Foundation funded a grant
to develop an infrastructure for bioengineering in the
United States. The AEMB, jointly with the U.S. National
Committee on Biomechanics (USNCB), was to develop a
unifying organization for bioengineering in the United
States. The co-principal investigators were Robert M.
Nerem and Arthur T. Johnson, and Patricia Horner served
as Project Director. The AEMB/USNCB Steering Committee consisted of Robert M. Nerem, Arthur T. Johnson,
Michael J. Ackerman, Gilbert B. Devey, Clifford E. Brubaker, Morton H. Friedman, Dov Jaron, Winfred M. Phillips, Alfred R. Potvin, Jerome S. Schultz, and Savio L-Y
Woo. The Steering Committee met in January and
March 1989, and the first workshop was held in August
1989. Two more Steering Committee meetings were held
in December 1989 and March 1990, and the second workshop was held in July 1990. The outcome of these two
workshops was to establish the American Institute for
Medical and Biological Engineering (AIMBE). All AEMB
members voted to cease operation of the Alliance for
Engineering in Medicine and Biology in 1990 and to
transfer the AEMB assets and 501(c)3 status to AIMBE
in 1991.
AIMBE opened an office in Washington, D.C., in 1995
with Kevin OConnor as Executive Director. He was succeeded by Arthur T. Johnson in 2004 and Patricia FordRoegner in 2005. AIMBE Presidents have been as follows:
Robert Nerem (19921993), Pierre Galletti (1994), Jerome
Schultz (1995), Winfred Phillips (1996), Larry McIntire
(1997), William Hendee (1998), John Linehan (1999),
Shu Chien (2000), Peer Portner (2001), Buddy Ratner
(2002), Arthur Coury (2003), Don Giddens (2004), Thomas
Harris (2005), and Herbert Voigt (2006).
Representing over 75,000 bioengineers, the AIMBE
seeks to serve and coordinate a broad constituency of
medical and biological scientists and practitioners, scientific and engineering societies, academic departments, and
industries. Practical engagement of medical and biological
engineers within the AIMBE ranges from the fields of

clinical medicine to food, agriculture, and environmental
bioremediation.
AIMBEs mission is to
Promote awareness of the field and its contributions
to society in terms of new technologies that improve
medical care and produce more and higher quality
food for people throughout the world.
Work with lawmakers, government agencies, and
other professional groups to promote public policies
that further advancements in the field.
Strive to improve intersociety relations and cooperation within the field.
Promote the national interest in science, engineering,
and education.
Recognize individual and group achievements
and contributions to medical and biological engineering.
AIMBE is composed of four sections:

The College of Fellows1000 Persons who are the
outstanding bioengineers in academic, industry, and
government. These leaders in the field have distinguished themselves through their contributions in
research, industrial practice, and/or education. Most
Fellows come from the United States, but there are
international Fellows.
The Academic CouncilUniversities with educational programs in bioengineering at the graduate
or undergraduate level. Currently there are approximately 85 member institutions. Representative to the
Council generally are chairs of their departments.
Many also are members of the College of Fellows.
The Council considers issues ranging from curricular
standards and accreditation to employment of graduates and funding for graduate study.
The Council of SocietiesThe AIMBEs mechanism
coordinating interaction among 19 scientific organizations in medical and biological engineering. The
purposes of the Council are to provide a collaborative
forum for the establishment of society member positions on issues affecting the field of medical and
biological engineering, to foster intersociety dialoge
and cooperation that provides a cohesive public representation for medical and biological engineering, and
to provide a way to coordinate activities of member
societies with the activities of academia, government,
the health-care sector, industry, and the public and
private biomedical communities.
The Industry CouncilA forum for dialog among
industry, academia, and government to identify
and act on common interests that will advance the
field of medical and biological engineering and contribute to public health and welfare. Industrial organizations may be members of the Industry Council if
they have substantial and continuing professional
interest in the field of medical and biological engineering.

Current members of the Council of Societies are as follows:
American College of Clinical Engineering
American Institute of Chemical Engineers; Food, Pharmaceutical and Bioengineering Division
American Medical Informatics Association
American Society of Agricultural and Biological Engineers
American Society for Artificial Internal Organs
American Society for Biomechanics
American Society of Mechanical Engineers, Bioengineering Division
Biomedical Engineering Society
Controlled Release Society
IEEE Engineering in Medicine and Biology Society
Institute of Biological Engineering
International Society for Magnetic Resonance in
Medicine
Orthopaedic Research Society
Rehabilitation Engineering and Assistive Technology
Society of North America
Society for Biomaterials
SPIE: The International Society for Optical Engineering
Surfaces in Biomaterials Foundation
Current members of the Industry Council are as follows:
Biomet, Inc.
Boston Scientific Corporation
Genzyme Corporation
Medtronic, Inc.
Pequot Ventures
Smith Nephew
Vyteris, Inc.
Wright Medical Technology, Inc.
Zimmer, Inc.
The AIMBE Board of Directors oversees the work of the
College of Fellows and the three councils. The Board consists of a President who is assisted by two Past Presidents,
the President-Elect, four Vice-Presidents at Large, a Secretary-Treasurer, and the Chair of the College of Fellows
all of whom are elected by the Fellows. The Board also
includes chairs of the other councils and chairs of all
standing committees. AIMBEs day-to-day operations are
supervised by the Executive Director in the Washington
headquarters.
AIMBEs Annual Event each winter in Washington,
D.C., provides a forum on the organizations activities
and is a showcase for key developments in medical and
315
biological engineering. The annual event includes a 1-day

scientific symposium sponsored by the College of Fellows, a
ceremony to induct the newly elected Fellows, and a 1-day
series of business meetings focused on public policy and
other issues of interest to AIMBEs constituents. For additional information about AIMBEs mission, memberships,
and accomplishments, visit http://www.aimbe.org.
The AIMBE has focused on public policy issues associated with medical and biological engineering. The
AIMBE enjoys high credibility and respect based on the
stature of its Fellows, support from constituent societies,
and its intention to be a forum for the best interests of the
entire field. The AIMBE has taken positions on several
important issues and advocated that they be adopted by
various agencies and by Congress. A few of the AIMBES
public policy initiatives that have met with success are as
follows:
National Institute of Biomedical Imaging and Bioengineering (NIBIB)Created in 2000 with the help of
AIMBE advocacy, the NIBIB has received strong
support from the AIMBE and other institutions that
value the role of technology in medicine, particularly
the Academy of Radiological Research. The NIBIB
has experienced rapid growth and development in all
areas, including scientific programs, science administration, and operational infrastructure. The prognosis for the near future is continued growth and
development especially in bioengineering, imaging,
and interdisciplinary biomedical research and training programs.
FDA Modernization Act (FDAMA)Enacted in 1997,
this legislation amended the Federal Food, Drug, and
Cosmetic Act relation to the regulation of food, drugs,
devices, and biological products. FDAMA enhanced
the FDAs mission in ways that recognized the Agency
would be operating in a twenty-first century characterized by increasing technological, trade, and public
health complexities.
Biomaterials Access Assurance ActThe 1998 legislation provides relief for materials suppliers to manufacturers of implanted medical devices by allowing
those suppliers to be dismissed from lawsuits in which
they are named if they meet the statutory definition of
a biomaterials supplier.
National Institutes of Health Bioengineering Consortium (BECON)This is the focus of bioengineering
activities at the NIH. The Consortium consists of
senior-level representatives from all NIH institutes,
centers, and divisions plus representatives of other
Federal agencies concerned with biomedical research
and development. The BECON is administered by
NIBIB.
The AIMBE Hall of Fame was established in 2005 to
recognize and celebrate the most important medical and
biological engineering achievements contributing to the
quality of life. The Hall of Fame provides tangible evidence
of the contributions of medical and biological engineering
during the following decades:
316
1. 1950s and earlier

Artificial kidney
X ray
Cardiac pacemaker
Cardiopulmonary bypass
Antibiotic production technology
Defibrillator
2. 1960s
Heart valve replacement
Intraocular lens
Ultrasound
Vascular grafts
Blood analysis and processing
3. 1970s
Computer-assisted tomography (CT)
Artificial hip and knee replacement
industry perspective (such as AdvaMed for the medical

device industry and the Biotechnology Industry Organization (BID) for the biotech industry), and there are peripheral
organizations that deal with public health, the environment,
and biotechnology. Many of these organizations publish
excellent journals, newsletters, and information sheets.
Those from trade organizations are often distributed free
of charge, but they do not include peer-reviewed articles.
Information about these can be found on the Internet.
Internationally, an organizational hierarchy exists.
National and transnational organizations can belong to
the International Federation for Medical and Biological
Engineering (IFMBE), and that confers membership
privileges to all AIMBE members and constituent society
members. The IFMBE and the International Organization
for Medical Physics (IOMP) together jointly sponsor a
World Congress on Medical Physics and Biomedical Engineering every 3 years. The IOMP and IFMBE are members
of the International Union for Physical and Engineering
Sciences in Medicine (IUPESM), and the IUPESM, in
turn, is a member of the International Council for Science
(ICSU). ICSU members are national and international
scientific unions and have a very broad and global outreach.
Balloon catheter
Endoscopy
THE FUTURE
Biological plant/food engineering

4. 1980s
Magnetic resonance imaging (MRI)
Laser surgery
Vascular stents
Recombinant therapeutics
5. 1990s
Genomic sequencing and micro-arrays
Positron emission tomography
Image-guided surgery
The AIMBE has now turned its attention to Barriers to

Further Innovation. It is providing forums and platforms
for identification and discussion of obstacles standing in
the way of advances in medical and biological engineering.
Barriers could be procedures, policies, attitudes, or information and education, anything that can yield when
AIMBE constituents apply pressure at appropriate levels.
OTHER SOCIETIES
These are other general biomedical engineering societies
that operate within the United States Among these, the
Biomedical Engineering Society (BMES), Engineering in
Medicine and Biology Society (EMBS), and the Institute
for Biological Engineering (IBE) are probably the most
inclusive. Others direct their attentions to specific parts of
the discipline. There are trade organizations that have an
At least for the foreseeable future, new groups will be

formed representing medical engineering specialties.
Whether these groups organize formally and persist will
depend on the continuing importance of their areas of
focus. The organizations with a more general foci will
continue to function and may spawn splinter groups. Given
the political importance of concerted effort, organizations
such as the AIMBE will continue to be active in promoting
policy. Competitive pressures among different organizations, especially when expectations of continuing growth
cannot be sustained, will always be a threat to the current
order. Given that the cycle of competition and disorder
leading to a realization that some ordered structure is
preferable has been repeated at least once, there will
continue to be some undercurrent of turmoil within the
community of medical engineering organizations.
U.S. PROFESSIONAL SOCIETIES AND ORGANIZATIONS
Biomedical Engineering Associations and Societies
Advamed. 1200 G Street NW, Suite 400, Washington,
D.C. 20005. 202-783-8700, http://www.advamed.org. Stephen J. Ubl, President. 1300 Members.
Represents manufacturers of medical devices,
diagnostic products, and medical information systems.
AdvaMeds members manufacture nearly 90% of the $80
billion of health-care technology purchased annually in the
United States. Provides advocacy, information, education,
and solutions necessary for success in a world of increasingly complex medical regulations.
Alpha Eta Mu Beta. 8401 Corporate Drive, Suite 140,
Landover, MD 20785. 301-459-1999, http://www.ahmb.org.
Herbert F. Voigt, National President; Patricia I. Horner,

Executive Director. 20 chapters.
Alpha Eta Mu Beta, the National Biomedical
Engineering Honor Society, was founded by Daniel Reneau
at Louisiana Tech University in 1979. This organization
was sponsored by the AEMB. The AEMB was established
to mark in an outstanding manner those biomedical engineering students who manifested a deep interest and
marked ability in their chosen life work to promote an
understanding of their profession and to develop its members professionally.
American Institute for Medical and Biological Engineering. 1901 Pennsylvania Ave NW, Suite 401, Washington,
D.C. 20006. 202-496-9660, http://www.aimbe.org. Patricia
Ford Roegner, Executive Director. 1000 Fellows; 18 Scientific Organizations; 85 Universities.
Founded in 1991 to establish an identity for the field of
medical and biological engineering, which is the bridge
between the principles of engineering science and practice
and the problems and issues of biological and medical
science and practice. The AIMBE comprises four sections.
The College of Fellows with over 1000 persons who are the
outstanding bioengineers in academia, industry, and government. The Academic Council is 85 universities with
educational programs in bioengineering at the graduate or
undergraduate level. The Council of Societies is 18 scientific organizations in medical and biological engineering.
The Industry Council is a forum for dialog among industry,
academia, and government. Principal activities include
participation in formulation of public policy, dissemination
of information, and education. Affiliated with the International Federation for Medical and Biological Engineering.
Annual event each winter in Washington, D.C.
American Association of Engineering Societies. 1828 L
Street NW, Suite 906, Washington, D.C. 20036. 202-2962237, http://www.aaes.org. Thomas J. Price, Executive
Director. 26 Engineering Societies.
Founded in 1979 in New York City. Member societies
represent the mainstream of U.S. engineering with more
than one million engineers in industry, government, and
academia. The AAES has four primary programs: communications, engineering workforce commission, international, and public policy. Governance consists of two
representatives from each of 26 member societies. Convenes diversity summits, publishes engineering and technology degrees, and holds annual awards ceremony.
American Academy of Environmental Engineers. 130
Holiday Court, Suite 100, Annapolis, MD 21401. 410266-3311, http://www.aaee.net. David A. Asselin, Executive Director.
The American Sanitary Engineering Intersociety Board
incorporated in 1955 became the American Academy of
Environmental Engineers in 1966; and in 1973, it merged
with the Engineering Intersociety Board. Principal purposes are improving the practice, elevating the standards,
and advancing public recognition of environmental engineering through a program of specialty certification of
qualified engineers.
317
American Academy of Orthopaedic Surgeons. 600 North

River Road, Rosemont, IL 60018. 847-823-7186, http://
www.aaos.org. Karen L. Hackett, Chief Executive Officer.
24,000 Members.
Founded in Chicago in 1933. Provides education and
practice management services for orthopedic surgeons and
allied health professionals. Maintains a Washington, D.C.,
office. Annual spring meeting.
American Academy of Orthotists and Prosthetists. 526
King Street, Suite 201, Alexandria, VA 22314. 703-8360788, http://www.oandp.org. Peter D. Rosenstein, Executive Director. 3000 Members.
Founded in 1970 to further the scientific and educational attainments of professional practitioners in the disciplines of orthotics and prosthetics. Members have been
certified by the American Board for Certification in Orthotics and Prosthetics. Annual spring meeting.
American Association of Physicists in Medicine. One
Physics Ellipse, College Park, MD 20740. 301-209-3350,
http://www.aapm.org. Angela R. Keyser, Executive Director. 4700 Members.
Founded in Chicago in 1958 and incorporated in
Washington in 1965. Promotes the application of physics
to medicine and biology. Member society of the American
Institute of Physics. Annual summer meeting.
American Chemical Society. 1155 Sixteenth Street NW,
Washington, D.C. 20036. 800-227-5558, http://www.chemistry.org. Madeleine Jacobs, Executive Director, and
CEO. 159,000 Members.
Founded in 1877 in New York City. Granted a national
charter by Congress in 1937. Encourages the advancement
of chemistry. Semiannual spring and fall meetings.
American College of Nuclear Physicians. 1850 Samuel
Morse Drive, Reston, VA 20190. 703-326-1190, http://
www.acnponline.org. Virginia M. Pappas, Executive Director. 500 Members.
Established in 1974, the organization provides access to
activities that encompass the business and economics of
nuclear medicine as they impact nuclear medicine physicians. Semiannual meetings fall and winter.
American College of Physicians. 190 N. Independence
Mall West, Philadelphia, PA 19106. 215-351-2600, http://
www.acponline.org. John Tooker, Executive Vice President. 119,000 Members.
Founded in New York City in 1915. Merged with the
Congress of Internal Medicine in 1923 and merged in 1998
with the American Society of Internal Medicine. Patterned
after Englands Royal College of Physicians. Members are
physicians in general internal medicine and related subspecialties. Maintains a Washington, D.C., office. Annual
spring meeting.
American College of Radiology. 1891 Preston White
Drive, Reston, VA 20191. 703-648-8900, http://www.
acr.org. Harvey L. Neiman, Executive Director. 30,000
Members.
318
Founded in 1923 in San Francisco and incorporated in

California in 1924. Purpose is to improve the health of
patients and society by maximizing the value of radiology
and radiologists by advancing the science, improving
patient service, and continuing education. Annual fall
meeting.
American College of Surgeons. 633 N. St. Clair Street,
Chicago, IL 60611. 312-202-5000, http://www.facs.org.
Thomas R. Russell, Executive Director. 64,000 Fellows,
5000 Associate Fellows.
Founded in 1913 and incorporated in Illinois. U.S.
member of the International Federation of Surgical Colleges. Members are Fellows who must meet high standards
established by the College. Purpose is to improve the
quality of care for the surgical patient by setting high
standards for surgical education and practice. Annual fall
clinical congress.
American Congress of Rehabilitation Medicine. 6801
Lake Plaza Drive, Suite B-205, Indianapolis, IN 46220.
317-915-2250, http://www.acrm.org. Richard D. Morgan,
Executive Director. 1700 Members, 15 Companies.
Founded in 1923 as the American College of Radiology
and Physiotherapy. Name changed in 1926 to American
College of Physical Therapy and in 1930 to American
Congress of Physical Therapy. Changed again in 1945 to
American Congress of Physical Therapy and in 1953
became American Congress of Physical Medicine and
Rehabilitation. Adopted its current name in 1967. Provides
education for professionals in medical rehabilitation. Fall
annual meeting.
American Institute of Biological Sciences. 1444 I Street
NW, Suite 200, Washington, D.C. 20005. 202-628-1500,
http://www.aibs.org. Richard OGrady, Executive Director.
80 Societies, 6000 Members.
Founded in 1947 as part of the National Academy of
Sciences. Incorporated as an independent nonprofit since
1954. Absorbed America Society of Professional Biologists
in 1969. Represents more than 80 professional societies
with combined membership exceeding 240,000 scientists
and educators. Also more than 6000 individual members.
Purpose is to better serve science and society. Annual
meeting in August.
American Institute of Chemical Engineers. 3 Park Avenue, New York, NY 10016. 212-591-8100, http://www.aiche.org. John Sofranko, Executive Director. 40,000
Members.
Organized in 1908 and incorporated in 1910. Member of
Accreditation Board for Engineering and Technology,
American National Standards Institute, American Association of Engineering Societies, and other related organizations. Purpose is to advance the chemical engineering
profession. Annual meeting in November.
American Institute of Physics. One Physics Ellipse,
College Park, MD 20740. 301-209-3131, http://www.aip.org. Marc H. Brodsky, Executive Director, and Chief
Executive Officer. 10 Societies and 24 Affiliates.
Chartered in 1931 to promote the advancement of physics and its application to human welfare. Federation of 10
Member Societies representing spectrum of physical
sciences.
American Institute of Ultrasound in Medicine. 14750
Sweitzer Lane, Suite 100, Laurel, MD 20707. 301-4984100, http://www.aium.org. Carmine Valente, Chief
Executive Officer.
Began in 1951 at a meeting of 24 physicians attending
the American Congress of Physical Medicine and Rehabilitation. Membership includes biologists, physicians, and
engineers concerned with the use of ultrasound for diagnostic purposes. Provides continuing education, CME
tests, and accreditation of ultrasound laboratories. Annual
fall meeting.
American Medical Informatics Association. 4915 St.
Elmo Avenue, Suite 401, Bethesda, MD 20814. 301-6571291, http://www.amia.org. Don Detmer, President, and
CEO. 3000 Members.
Founded in 1990 through a merger of three existing
health informatics associations. Members represent all
basic, applied, and clinical interests in health-care information technology. Promotes the use of computers and
information systems in health care with emphasis on direct
patient care. Semiannual meetings: spring congress in the
West and fall annual symposium in the East.
American Society for Artificial Internal Organs. P.O. Box
C, Boca Raton, FL 33429. 561-391-8589, http://www.asaio.
net. 1400 Members.
Established in 1955 in Atlantic City, NJ. Annual June
conference.
American Society for Engineering Education. 1818 N
Street NW, Suite 600, Washington, D.C. 20036. 202331-3500, http://www.asee.org. Frank L. Huband, Executive Director. 12,000 Members, 400 Colleges, 50 Corporations.
Founded in 1893 as the Society for Promotion of Engineering Education. Incorporated in 1943 and merged in
1946 with Engineering College Research Association.
Members include deans, department heads, faculty members, students, and government and industry representatives from all disciplines of engineering and engineering
technology. Member of the American Association of Engineering Societies, Accreditation Board for Engineering
and Technology, American Institute for Medical and
Biological Engineering, and American Council on
Education. Participating society of World Federation of
Engineering Associations. Purpose is to further education
in engineering and engineering technology. Annual June
meeting.
American Society for Healthcare Engineering of the American Hospital Association. One North Franklin, 28th Floor,
Chicago, IL 60606. 312-422-3800, http://www.ashe.org.
Albert J. Sunseri, Executive Director.
Affiliate of the American Hospital Association. Annual
June meeting.
American Society for Laser Medicine and Surgery. 2404

Stewart Avenue, Wausau, WI 54401. 715-845-9283, http://
www.aslms.org. Dianne Dalsky, Executive Director. 3000
Members.
Founded in 1980 to promote excellence in patient care
by advancing laser applications and related technologies.
Annual spring meeting.
American Society of Agricultural and Biological Engineers. 2950 Niles Road, St. Joseph, MI 49085. 269-4290300, http://www.asabe.org. Melissa Moore, Executive Vice
President. 9000 Members.
Founded in 1907 as the American Society of Agricultural Engineers and changed its name in 2005. Dedicated
to the advancement of engineering applicable to agricultural, food, and biological systems. Annual meeting in
July.
American Society of Civil Engineers. 1801 Alexander
Bell Drive, Reston, VA 20191. 703-295-6000, http://
www.asce.org. Patrick J. Natale, Executive Director.
137,500 Members.
Founded in 1852 as the American Society of Civil Engineers and Architects. Dormant from 1855 to 1867, but it
revived in 1868 and incorporated in 1877 as the American
Society of Civil Engineers. Over 400 local affiliates, 4
Younger Member Councils, 230 Student Chapters, 36 Student Clubs, and 6 International Student Groups. Semiannual spring and fall meetings.
American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. 1791 Tullie Circle NE, Atlanta, GA
30329. 404-636-8400, http://www.ashrae.org.
Incorporated in 1895 as the American Society of Heating
and Ventilating Engineers, known after 1954 as American
Society of Heating and Air-Conditioning Engineers.
Merged in 1959 with American Society of Refrigerating
Engineers to form American Society of Heating, Refrigerating and Air-Conditioning Engineers. Annual summer
meeting.
American Society of Mechanical Engineers. Three Park
Avenue, New York, NY 10016. 212-591-7722, http://
www.asme.org. Virgil R. Carter, Executive Director.
120,000 Members.
Founded in 1880 and incorporated in 1881. Focuses on
technical, educational, and research issues of engineering
and technology. Sets industrial and manufacturing codes
and standards that enhance public safety. Conducts one of
the worlds largest technical publishing operations. Semiannual summer and winter meetings.
American Society of Neuroradiology. 2210 Midwest
Road, Suite 207, Oak Brook, IL 60523. 630-574-0220,
http://www.asnr.org. James B. Gantenberg, Executive
Director/CEO. 2700 Members.
Founded in 1962. Supports standards for training in the
practice of neuroradiology. Annual spring meeting.
American Society of Safety Engineers. 1800 E. Oakton
Street, Des Plaines, IL 60018. 847-699-2929, http://
319
www.asse.org. Fred Fortman, Executive Director. 30,000

Members.
Founded in 1911 as the United Association of Casualty
Inspectors and merged with the National Safety Council in
1924, becoming its engineering section. Became independent again in 1947 as the American Society of Safety
Engineers and incorporated in 1962. There are 13 practice
specialties, 150 chapters, 56 sections, and 64 student sections. Annual spring meeting.
Association for Computing Machinery. 1515 Broadway,
New York, NY 10036. 212-626-0500, http://www.acm.org.
John R. White, Executive Director. 80,000 Members.
Founded in 1947 at Columbia University as Eastern
Association for Computing Machinery and incorporated in
Delaware in 1954. Affiliated with American Association for
Advancement of Science, American Federation of Information Processing Societies, Conference Board of Mathematical Sciences, National Academy of Sciences-National
Research Council, and American National Standards Institute. Advancing the skills of information technology professionals and students. Annual fall meeting.
Association for the Advancement of Medical Instrumentation. 1100 North Glebe Road, Suite 220, Arlington, VA
22201. 703-525-4890, http://www.aami.org. Michael J.
Miller, President. 6000 Members.
Founded in 1967, the AAMI is an alliance of over 6000
members united by the common goal of increasing the
understanding and beneficial use of medical instrumentation and technology. Annual spring meeting.
Association of Biomedical Communications Directors.
State University of New York at Stony Brook, Media
Services L3044 Health Sciences Center, Stony Brook,
NY 11794. 631-444-3228. Kathleen Gebhart, Association
Secretary. 100 Members.
Formed in 1974 as a forum for sharing information;
adopted a Constitution and Bylaws in 1979, and incorporated in April 1979 in North Carolina. Members are directors of biomedical communication in academic health
science settings. Annual spring meeting.
Association of Environmental Engineering and Science
Professors. 2303 Naples Court, Champaign, IL 61822.
217-398-6969, http://www.aeesp.org. Joanne Fetzner,
Secretary. 700 Members.
Formerly, in 1972, the American Association of Professors in Sanitary Engineering. Professors in academic programs throughout the world who provide education in the
sciences and technologies of environmental protection.
Biennial conference in July.
Biomedical Engineering Society. 8401 Corporate Drive,
Suite 140, Landover, MD 20785. 301-459-1999, http://
www.bmes.org. Patricia I. Horner, Executive Director.
3700 Members.
Founded in 1968 in response to a need to give equal
representation to both biomedical and engineering interests. The purpose of the Society is to promote the increase
of biomedical engineering knowledge and its use. Member
320
of American Institute for Medical and Biological Engineering. Annual fall meeting.
Biophysical Society. 9650 Rockville Pike, Bethesda,
MD 20814. 301-634-7114, http://www.biophysics.org. Ro
Kampman, Executive Officer. 7000 Members.
Founded in 1957 in Columbus, OH, to encourage development and dissemination of knowledge in biophysics.
Annual winter meeting.
Health Physics Society. 1313 Dolley Madison Boulevard, Suite 402, McLean, VA 22101. 703-790-1745, http://
www.hps.org. Richard J. Burk, Jr, Executive Secretary.
7000 Members.
Founded in 1956 in the District of Columbia and reincorporated in Tennessee in 1969. Society specializes in
occupational and environmental radiation safety.
Affiliated with International Radiation Protection Association. Annual summer meeting.
Human Factors and Ergonomics Society. P.O. Box 1369,
Santa Monica, CA 90406. 310-394-1811, http://www.
hfes.org. Lynn Strother, Executive Director. 50 Active
Chapters and 22 Technical Groups.
Founded in 1957, formerly known as the Human Factors Society. An interdisciplinary organization of professional people involved in the human factors field. Member
of the International Ergonomics Association. Annual meeting in September-October.
Institute for Medical Technology Innovation. 1319 F
Street NW, Suite 900, Washington, D.C. 20004. 202-7830940, http://www.innovate.org. Martyn W.C. Howgill,
Executive Director.
The concept was developed in 2003 by leaders in the
medical device industry, and the Institute was incorporated and opened its offices in 2004. Purpose is to demonstrate the role, impact, and value of medical technology on
health care, economy, and society, for the benefit of
patients.
Institute of Electrical and Electronics Engineers. 3 Park
Avenue, 17th Floor, New York, NY 10016. 212-419-7900,
http://www.ieee.org. Daniel J. Senese, Executive Director.
365,000 Members.
The American Institute of Electrical Engineers was
founded in 1884 and merged in 1963 with the Institute
of Radio Engineers. Three Technical Councils, 300 local
organizations, 1300 student branches at universities, and
39 IEEE Societies including the Engineering in
Medicine and Biology Society with 8000 members and
meets annually in the fall. Maintains Washington, D.C.
office.
Institute of Environmental Sciences and Technology. 5005
Newport Drive Suite 506, Rolling Meadows, IL 60008. 847255-1561, http://www.iest.org. Julie Kendrick, Executive
Director.
Formed by a merger of the Institute of Environmental
Engineers and the Society of Environmental Engineers in
1953. Annual spring meeting.
Instrument Society of America. 67 Alexander Drive, Research Triangle Park, NC 27709. 919-549-8411, http://www.
isa.org. Rob Renner, Executive Director. 30,000 Members.
Founded in Pittsburgh in 1945. Charter member of
American Automatic Control Council, affiliate of American Institute of Physics, member of American Federation
of Information Processing Societies, member of American
National Standards Institute, and U.S. representative to
the International Measurement Confederation. Develops
standards, certifies industry professionals, provides education and training, publishes books and technical articles, and hosts largest conference for automation
professionals in the Western Hemisphere. Annual meeting in October.
International Biometric Society ENAR. 12100 Sunset
Hills Road, Suite 130, Reston, VA 22090. 703-437-4377,
http://www.enar.org. Kathy Hoskins, Executive Director.
6500 Members.
Founded in September 1947. Became the International
Biometric Society in 1994. Annual March and June meetings.
International College of Surgeons. 1516 North Lake
Shore Drive, Chicago, IL 60610. 312-642-3555, http://
www.icsglobal.org. Max C. Downham, Executive Director.
10,000 Members.
Founded in Geneva, Switzerland, in 1935 and incorporated in the District of Columbia in 1940. Federation of
general surgeons and surgical specialists. Annual spring
meeting of U.S. section and biennial international meetings.
International Society for Magnetic Resonance in Medicine. 2118 Milvia Street, Suite 201, Berkeley, CA 94704.
510-841-1899, http://www.ismrm.org. Roberta A. Kravitz,
Executive Director. 6,000 Members.
Formed as a merger of the Society for Magnetic Resonance Imaging and Society of Magnetic Resonance in
Medicine in 1995. Promotes the application of magnetic
resonance techniques to medicine and biology. Annual
meetings in April/May.
Medical Device Manufacturers Association. 1919 Pennsylvania Avenue NW, Suite 660, Washington, D.C. 20006.
202-349-7171, http://www.medicaldevices.org. Mark B.
Leahey, Executive Director. 140 Companies.
Created in 1992. Supersedes Smaller Manufacturers
Medical Device Association. Represents manufacturers
of medical devices. Annual May meeting.
Radiation Research Society. 810 East 10th Street,
Lawrence, KS 666044. 800-627-0629, http://www.
radres.org. Becky Noordsy, Executive Director. 2025
Members.
Founded in 1952 as a professional society of persons
studying radiation and its effects. Annual spring-summer
meetings.
Radiological Society of North America. 820 Jorie Boulevard, Oak Brook, IL 60523. 630-571-2670, http://
www.rsna.org. Dave Fellers, Executive Director. 37,577

Members.
Founded as Western Roentgen Society and assumed its
current name in 1918. Members are interested in the
application of radiology to medicine. Holds the largest
medical meeting in the world annually in November with
more than 60,000 in attendance.
RESNARehabilitation Engineering & Assistive Technology Society of North America. 1700 N. Moore Street, Suite
1540, Arlington, VA 22209. 703-524-6686, http://www.resna.org. Larry Pencak, Executive Director. 1000 Members.
Founded in 1979 as the Rehabilitation Engineering
Society of North America. In June 1995, the name was
changed to the Rehabilitation Engineering and Assistive
Technology Society of North AmericanRESNA. Twentyone special interest groups and seven professional specialty groups. Annual meeting in June.
SPIEInternational Society for Optical Engineering. P.O.
Box 10, Bellingham, WA 98227. 360-676-3290, http://
www.spie.org. Eugene G. Arthurs, Executive Director.
14,000 Members, 320 Companies.
Founded in 1956 in California as the Society of Photographic Instrumentation Engineers, it later became the
Society of Photo-Optical Instrumentation Engineers and
assumed its current name in 1981. Members are scientists,
engineers, and companies interested in application of optical, electro-optical, fiber-optic, laser, and photographic
instrumentation systems and technology. Semiannual
meetings.
Society for Biological Engineering of the American Institute
of Chemical Engineers. 3 Park Avenue, New York, NY
10016. 212-591-7616, http://www.bio.aiche.org.
Established by the AIChE for engineers and applied
scientists integrating biology with engineering.
Society for Biomaterials. 15000 Commerce Parkway,
Suite C, Mt. Laurel, NJ 08054. 856-439-0826, http://
www.biomaterials.org. Victoria Elliott, Executive Director.
2100 Members.
Founded in 1974. Promotes biomaterials and their
uses in medical and surgical devices. Annual spring
meeting.
Society for Biomolecular Screening. 36 Tamarack Avenue, #348, Danbury, CT 06811. 203-743-1336, http://
www.sbsonline.org. Christine Giordano, Executive Director. 1080 Members, 230 Companies.
Supports research in pharmaceutical biotechnology and
the agricultural industry that use chemical screening procedures. Annual fall meeting.
Society for Modeling and Simulation International. P.O.
Box 17900, San Diego, CA 92177. 858-277-3888, http://
www.scs.org. Steve Branch, Executive Director.
Established in 1952 as the Simulation Council and
incorporated in California in 1957 as the Simulation Councils. Became Society for Computer Simulation in 1973 and
later changed its name to the current one. A founding
321
member of the Information Processing Societies and

National Computer Confederation Board, and affiliated
with American Association for the Advancement of
Science. Holds regional simulation multiconferences.
Society of Interventional Radiology. 3975 Fair Ridge
Drive, Suite 400 North, Fairfax, VA 22033. 703-691-1805,
http://www.sirweb.org. Peter B. Lauer, Executive Director.
Society of Nuclear Medicine. 1850 Samuel Morse Drive,
Reston, VA 20190. 703-709-9000, http://www.interactive.snm.org. Virginia Pappas, Executive Director.
Society of Rheology. Suite 1N01, 2 Huntington Quadrangle, Melville, NY 11747. 5162403, http://www.rheology.org. Janis Bennett, Executive Director.
Permanent address is at the American Institute of
Physics and is one of five founding members of the AIP.
Members are chemists, physicists, biologists, and others
concerned with theory and precise measurement of flow of
matter and response of materials to mechanical force.
Annual meeting held in October or November.
Professional Societies and Organizations of Other Countries
Pick up from IFMBE Affiliates on www.ifmbe.org.
Further Reading
Biomedical Engineers, Brief 519, G.O.E. 02.02.01; D.O.T. (4th ed.)
019. Chronicle Guidance Publications, Moravia, NY 13118,
1994.
Biomedical Engineers. Occupational Outlook Handbook, 2004-05
Edition. Bureau of Labor Statistics, US. Department of Labor.
http://www.bls.gov/oco/ocos262.htm.
Boykin D. Biomedical engineering takes center stage. Engineering
Times 2004; 26 (9). National Society of Professional Engineers,
1420 King Street, Alexandria, VA 22314.
Collins CC. The retrospectroscope: Notes on the history of biomedical engineering in America. IEEE Eng Med Biol Mag Dec.
1988. IEEE Engineering in Medicine & Biology Society, 445
Hoes Lane, Piscataway, NJ 08854.
Enderle J, editor. Charting the milestones of biomedical engineering. IEEE Eng Med Biol Mag, May 2002. IEEE Engineering in
Medicine and Biology Society, 445 Hoes Lane, Piscataway, NJ
08854.
Fagette PH je, Homer PI, editor. The Biomedical Engineering
Society: An Historical Perspective. Landover, MD: The Biomedical Engineering Society; 2004.
Goodman L. The International Federation for Medical and Biological Engineering20 years on. Med Biol Eng Comput Jan.
1978. International Federation for Medical and Biological
Engineering.
Katona P. The Whitaker Foundation: The end will be just the
beginning. IEEE Trans Med Imaging 2002;21(8). IEEE Engineering in Medicine & Biology Society, 445 Hoes Lane, Piscataway, NJ 08854.
Johnson AT. Executive Directors Report: What is ASMBE all
about? ASMBE Newslett 2004:1.
Planning a Career in Biomedical Engineering (2004). Biomedical
Engineering Society, 8401 Corporate Drive, Suite 140, Landover, MD 20785. http://www.bmes.org
Tompkins W. From the President. IEEE Eng Med Biol Mag,
December 1988. IEEE Engineering in Medicine & Biology
Society, 445 Hoes Lane, Piscataway, NJ 08854.
322

TADEUSZ M. DRZEWIECKI
JERRY M. CALKINS
Defense Research Technologies,
Inc.
Rockville, Maryland
INTRODUCTION
Medical gas monitoring has been so successful in improving patient safety and reducing patient risk that it has
become standard medical practice today in every part of
medical practice from hospital to home. The argument for
providing additional patient safety will continue to be a
powerful incentive to improve and enhance the methods
and techniques to provide increased knowledge of the
monitoring of respiratory and anesthetic gases. Research
on gas markers to aid in diagnosis is an equally important
application, and with the capability to measure gases at
parts per billion or even trillion, heretofore unobserved
gases can point to early diagnosis of such nearly always
fatal neonatal diseases as necrotizing enterocolitis and
other difficult-to-diagnose states.
Medical gas analyzers are used to sample and measure
gases of medical importance, such as anesthesia and
respiratory monitoring, and detection of trace gases for
diagnostic purposes. This article predominantly discusses
these two cases. The estimation of arterial blood gases is
considered only in terms of measurement of respired gases.
Two basic categories of sensor/analyzers exist: continuous
and batch. Gas analyzers are further broken down by
their sensing mechanisms into two fundamental modes
of operation, specific and analytic.
Continuous devices are used where real-time information is needed. Batch systems operate on a bolus of gas,
usually when real-time information is not needed. Many
applications may have to live with the offline, longer
duration of a batch test because nothing else is available.
Specific-type sensors rely on particular physical phenomena that are activated in the presence of a particular
gas. Electrochemical devices, for example, are representative of a specific sensor wherein a voltage is developed by a
chemical reaction between the sensor material and the gas
being analyzed in some identified or known proportion to
the amount of gas present. The same is true of fuel cells and
other galvanic devices where an electric potential is developed in the presence of a difference in partial pressures
across a conducting medium.
Specificity is a major issue and is of particular importance to the medical community. At this point in time, no
truly specific sensors exist. All sensors exhibit some form of
cross-sensitivity to a variety of gases, some in the same
family, some with similar physical properties, and some for
extraneous reasons not always obvious to the user. Nitrous
oxide (N2O) and carbon dioxide (CO2) have practically the
same molecular weight, 44.0128 versus 44.0098. Consequently, they have near-identical physical characteristics
such as specific heat and viscosity. Interestingly, they also
have almost exactly the same absorption wavelengths,
although not necessarily because the atomic weights are
the same but rather because the orbital electron transition

energetics are similar. Thus, it is difficult to distinguish the
two with most conventional techniques, and a carbon
dioxide sensor that is based on absorption at 4.3 mm will
be affected by the presence of nitrous oxide with its peak
absorption at 4.5 mm. Unless a very narrow wavelength
light source is used, such as a laser, the nitrous oxide will
absorb some of the energy and make it appear that more
carbon dioxide is present than there really is.
Analytic devices imply an ability to assay a gas or gas
mixture and tell the user not only how much of a particular
gas is present, but also which gases or elements are present
and in what relative quantities, of which optical spectroscopy is a good example where a large number of absorption
lines exist for different gases, so one can scan the entire
spectrum from ultraviolet (UV) to far infrared (IR) and
compare absorption lines to see what is present. This example is interesting because IR spectroscopy can also be the
basis for a specific sensor when only a single or a particular
wavelength of light is monitored looking only for a gas at
that absorption line. Gas chromatography (GC) is also a
batch process where a bolus of gas to be assayed is separated
into its constituent parts in time by a molecular sieve and
the binary pairs (the carrier and the separated gas) are
detected and quantized upon exiting the GC column.
Perhaps the most common use of and need for continuous gas analysis or sensing is in real-time respiratory
applications where inspired and expired (end-tidal) concentrations of respiratory gases are measured to validate
that the appropriate standards of care are being applied
and that proper ventilation and oxygenation of a patient is
being achieved. An example would be monitoring of a
ventilated patient in the ICU or recovery room. In addition,
the monitoring of anesthetic gases during and immediately
following anesthetic administration in the operating room
is a critical application that can mean the difference
between life and death. Too much can lead to brain damage
or death, and too little can result in unnecessary pain and
memory recall. And, of course, the detection and warning of
the presence of toxic gases such as carbon monoxide
released from desiccated soda lime CO2 scrubbers due to
interactions between the anesthetic agents and various
scrubbers, or the production of the highly toxic Compound
A, is critical to patient safety.
Continuous medical gas monitoring provides the
clinician with information about the patients physiologic
status, estimates of arterial blood gases, verifies that the
appropriate concentrations of delivered gases are administered, and warns of equipment failure or abnormalities in
the gas delivery system. Monitors display inspired and
expired gas concentrations and sound alarms to alert
clinical personnel when the concentration of oxygen (O2),
carbon dioxide (CO2), nitrous oxide (N2O), or volatile anesthetic agent falls outside the desired set limits.
Medical gas analysis has been driven by a need for
safety and patient risk reduction through respiratory
gas analysis and identification and quantification of volatile anesthetic vapors. Perhaps one of the earliest anesthetic agent sensors was the Drager Narkotest, which
comprised a polymer rubber membrane that contracted
and moved a needle as it absorbed agent. It did not require
an electrical power supply and its slow response was not

any slower than the rate of change of gas composition.
Much has transpired since those early days.
Currently, numerous methods and techniques of gas
monitoring are in place, and new techniques and paradigms are constantly being developed to meet a new need or
to serve the community with better performance or lower
cost. In this review, many of the intrinsic advantages and
disadvantages of these methods and techniques are discussed. A brief comparison, which includes stand-alone
and multioperating room gas monitors that can determine
concentrations of anesthetic and respiratory gases in
the patient breathing circuit during anesthesia, is also
presented.
Much of the research and development of these monitors
have followed the long use of similar detector principles
from analytical chemistry. As a result of the fast pace of
sensor development, to a great extent driven by the need
for hazardous gas sensors in the face of terrorist threats
and being spearheaded by agencies such as the Defense
Departments Defense Advanced Research Projects Agency
(DARPA), an attempt is made to cover the most common
systems and provide insights into the future based on solid
technological developments.
The current development of gas analyzers is described
in the extensive anesthesia and biomedical engineering
literature. Complete and specific historical information
about the principles and applications of these devices is
well reviewed in several texts [e.g., Ref. (1)], manufacturers and trade publications [(2) (ECRI)], and an extensive open literature describing equipment and operating
principles, methods, and techniques that is available on the
Internet. Societies and professional associations also exist
that deal with just one method of gas analysis that can
provide in-depth information about their particular
interests. The Chromatographic Society is one such organization. It is the purpose of this article to concisely summarize such a large selection of information sources to a
manageable few, but with enough references and pointers
to allow even the casual reader to obtain whatever relevant
information at whatever level is required.
CURRENT GAS MONITOR METHODS AND TECHNIQUES
As a result of the chemically diverse substances to be
measured, medical gas analyzers commonly combine more
than one analytical method. Methods of interest to the
medical practitioner, clinician, researcher, or operator
include, in alphabetical order:
Colorimetry
Electrochemistry
Fuel cells
Polarography
Gas chromatography
Flame ionization
Photoionization Detectors
Thermal conductivity

323
Infrared/Optical Spectroscopy
Luminescence/fluorescence
Mass spectrometry
Nuclear Magnetic Resonance
Paramagnetism
Radioactive ionization
Raman Laser Spectroscopy
Solid-state sensors

Semiconductor metal oxides

ChemFETs
Solid-state galvanic cells
Piezoelectric/Surface Acoustic Wave
Each of these methods will be described in the following

text. Illustrative examples of typical devices may be
mentioned.
COLORIMETRY
Colorimetry is one of the oldest methods of gas analysis
that is typically used to detect the presence of carbon
dioxide in a breath as a means of determining if proper
tracheal intubation has been performed. Basically, it works
on the principle of changing the color of a material such as
paper or cloth impregnated with a reagent in the presence
of a known analyte. An example is the changes in litmus
paper from purple to red in the presence of an acid and blue
in the presence of a base. Carbon dioxide (CO2) in the
presence of water vapor in the exhaled breath produces
carbonic acid, which turns the litmus paper toward red,
usually some shade of pink. The degree of color change can
be quite subjective, but a color scale usually accompanies
most devices so that a coarse estimate, roughly 0.5% CO2
by volume, can be made. More expensive, sophisticated
devices offer an electronic colorimetric analyzer that does
the comparison automatically and can even provide a
digital output.
Reagents may be tuned to a variety of specific gases and
are quite commonly used in the semiconductor business to
monitor levels of hydride gases to include arsine and
phosphene. Hydrogen sulfide (H2S) is also a common
analyte for colorimetric sensors (1).
Cross-sensitivity can be additive or subtractive with
colorimetric devices. For example, a colorimetric capnometer (CO2 sensor) may register false-positives and
false-negatives. Color may change in the presence of acidic
reflux or the ingestion of acidic liquids (lemonade, wine,
etc.). Conversely, the presence of bases could negate the
acid response, which is true in other applications as well.
For example, in hydrogen sulfide detection, the presence of
methyl mercaptan (CH4S) compounds can cancel out any
reading of H2S. Clearly, any chemical reactions between
the selected analyte and a reactive contaminant can affect
readings one way or another.
Figure 1 shows a photograph of one of the newer colorimetric capnometers on the market manufactured by
Mercury Medical (www.mercurymed.com). A colorchanging tape used in this device turns yellow in the
324
Figure 1. Mercury medical colorimetric

end tidal CO2 detector.
presence of CO2, but returns to green when no CO2 is

present. The same piece of paper will register changes
as the patient breathes. As the tape is consumed, it can
be pulled through the device, and a fresh piece exposed to
the breath.
ELECTROCHEMISTRY
Electrochemical gas sensors operate on the principle that a
current is generated when the selected gas reacts at an
electrode in the presence of an electrolyte, not unlike a
battery. For this reason, these devices are often called
amperometric gas sensors or microfuel cells. Electrochemical gas sensors are found ubiquitously in industry
because of their excellent sensitivity to toxic gases (often
low parts per million, ppm) and relatively low cost. They
are, however, consumable devices and have a limited operating and shelf life, typically a few years. They are not
particularly susceptible to poisoning, that is, being contaminated or degraded by absorption of particular contaminant gas species. Gases of medical importance that
can be sensed with electrochemical devices are ammonia,
carbon monoxide, nitric oxide, oxygen, ozone, and sulfur
dioxide. The most prominent medical use is in the
measurement of oxygen.
Three basic elements exist in any electrochemical gas
sensor. The first is a gas-permeable, hydrophobic membrane, which allows gas to diffuse into the cell but keeps
the liquid or gel electrolyte inside. It is the slow diffusion
process that limits the time response of these sensors,
although, if they are made very small, they can be quite
responsive. Typically, however, an oxygen sensor may take
as long as 20 s to equilibrate, making these devices impractical for real-time monitoring of respiration other than
monitoring some average oxygen level.
The second element is the electrode. Selection of the
electrode is critical to the selectivity of an appropriate
reaction. Typically, electrodes are catalyzed noble metals
such as gold or platinum. Gases such as oxygen, nitrogen
oxides, and chlorine, which are electrochemically reducible, are sensed at the cathode while those that are electrochemically oxidizable, such as carbon monoxide, nitrogen
dioxide, and hydrogen sulfide, are sensed at the anode.
The third element is the electrolyte that carries the

ions between the electrodes. The electrolyte must be
kept encapsulated in the cell as leakage would cause
dysfunction.
In many cases, a fourth element exists which is a filter/
scrubber mounted across the face of the sensor and the
permeable membrane, which helps with the specificity by
eliminating some interfering gases. In an oxygen sensor,
this element is often an activated charcoal molecular sieve
that filters out all but carbon monoxide and hydrogen.
Other filters can be tailored to allow only the selected
analyte through.
An oxygen fuel cell gas detector uses a lead anode that is
oxidized during operation. It is powered by the oxygen it is
sensing with a voltage output proportional to the oxygen
concentration in the electrolyte. In this case, the electrolyte
is potassium hydroxide (KOH) solution. With the recent
explosion in research on fuel cells, they have become almost
ubiquitous in medical practice, supplanting the Clark electrodes to a great extent.
Among the most recognizable oxygen-sensing devices
are the Clark electrodes (Ag/AgCl anode, Pt cathode). One
of the first applications of this device was monitoring
oxygen concentrations in the inspiratory limb of the
breathing circuit of an anesthesia machine. Clark electrodes differ slightly from the above-described fuel-cell-type
devices. Clark electrodes require an external voltage
source to generate a bias voltage against which the
oxygen-induced potential operates. These devices are,
therefore, called polarographic because of the bias voltage
that is required for its operation, contrasting with a galvanic or amperometric cell, which produces current on its
own proportional to the amount of analyte present. Oxygen
from the sample fluid equilibrates across a Teflon membrane with a buffered potassium chloride (KCl) solution
surrounding a glass electrode. The electrode has a platinum cathode and a silver/silver chloride anode. With
between 0.5 V and 0.9 V applied across the electrodes,
the consumption of O2 at the cathode, and hence the
current in the circuit, is dependent on the O2 concentration
in the solution, which rapidly equilibrates with the sample.
In practice, 0.68 V is used. Performance is adversely
affected by the presence of N2O and halogenated anesthetic
agents such as halothane. Protection of the platinum
Figure 2. Datex-Ohmeda Smart Vent 7900 ventilator monitor

uses a galvanic cell O2 sensor.
cathode and the need for semipermeable membranes

reduces their effectiveness.
Fuel cell and polarographic devices both require temperature and pH compensation and, as indicated before,
have limited life spans because of the consumable nature of
the reaction and the propensity of the permeable membranes to eventually lose the effectiveness of the Clark
electrodes.
Datex-Ohmeda, one of the largest manufacturers
of anesthesia equipment, sells many of their systems,
included with which is a galvanometric fuel cell oxygen
sensor. Figure 2 shows a Smart Vent 7900TM (3) display
showing the level of oxygen being delivered. The oxygen
sensor life is specified at 18 months.
GAS CHROMATOGRAPHY
Gas chromatography (GC) is an analytic tool that provides
the user with an assay of what is in the gas sample of
interest. It consists of two parts: (1) separation of different
species by differential travel times through a separation
column and (2) analytic detection of the quantity of each
analyte at the end of the column using any one of a variety
of methods. That is, GC provides a quantification of the
constituent gases as well as an identification of what the
constituent gases are. It is actually a very simple process,
and, were it not for the relatively long analysis time,
usually on the order of several minutes to as long as
an hour, and its batch nature, it would be used more
frequently.
GC requires the separation of a gas sample into its
constituent gases, which is accomplished by mixing the
sample with a carrier gas, usually some inert gas like
helium or nitrogen, which is not adsorbed by the
GC column, and passing it through a column or long
impermeable, nonreacting (e.g., stainless steel) capillary
filled with a zeolite, molecular sieve, or other material that
separates the sample gases according to their physical or
chemical properties. The different gas constituents travel
through the column at different speeds and exit as binary
325
pairs (a constituent and the carrier) in order of their

adsorption properties. The time it takes for a constituent
to exit the column identifies the constituent. The capillary
columns can be as short as a few centimeters to tens and
even hundreds of meters long. The capillary columns are
heated to maintain a constant temperature, as the transport characteristics tend to be temperature-dependent.
Calibration is required to determine the transit times
for each analyte, which is done by injecting known gas
samples at the start of the column and physically timing
them at the exit. At the exit of the column, a gas detector
exists usually a flame ionization or thermal conductivity
detector that measures the amount of constituent relative
to the carrier. After all the constituents have been
accounted for, the assay of the original sample can be made
by summation of the relative amounts of each constituent
and then taking the ratio of each constituent relative to the
summation, which then gives the concentrations and the
final assay.
Usually, this summation is accomplished by summing
the areas under the detector output peaks and then ratioing the areas under the individual peaks relative to the
total area. The choice of gas chromatographic detectors
depends on the resolution and accuracy desired and
includes (roughly, in order from most common to the least):
the flame ionization detector (FID), thermal conductivity
detector (TCD or hot wire detector), electron capture detector (ECD), photoionization detector (PID), flame photometric detector (FPD), thermionic detector, and a few
more unusual or expensive choices like the atomic emission
detector (AED) and the ozone- or fluorine-induced chemiluminescence detectors.
The Flame Ionization Detector (FID) is widely used to
detect molecules with carbon-hydrogen bonds and has good
sensitivity to low ppm. Basically, in operation, the analyte
is injected into a hydrogen carrier and ignited inside a
grounded metal chamber. Hydrogen is used because it
burns clean and is carbon-free. The latter is important
because output is proportional to the ionized carbon atoms.
An electrode is situated just above the flame and a voltage
potential is applied. The current produced is proportional
to the number of carbon atoms in the analyte. When
applied at the exit of a GC, very accurate measures of
hydrocarbon gases can be made.
Photoionization Detectors (PIDs) are used to detect the
volatile organic compound (VOC) outputs of GCs but are
also widely used in industry and science to detect environmental and hazardous gases. They operate on a similar
principle to that of the FID, but use ultraviolet light (UV) as
opposed to a flame to ionize the flowing gas between
insulated electrodes. As UV energy is a much higher
frequency (lower wavelength) than IR or visible light, it
can be larger and, consequently, can readily ionize gases.
The ionization potentials of the analyte gases are matched
by adjusting the frequency of the emitted light. The output
power of the lamp is roughly the product of the number of
photons and the energy per photon divided by the area and
time, although changing the output frequency will change
the photon energy (E hv), thereby changing the power.
The output power can be changed independently by
increasing the fluence of photons.
326
Figure 3. Seito ToxiRae personal PID gas monitor.
An inert gas lamp provides the UV light, (e.g., xenon

lamps emit UV light at 147.6 nm, krypton at 123.9 nm, and
argon at 105.9 nm). An advantage is that the sensitivity to
particular species or groups of compounds can be adjusted
by adjusting the output power to match the distinct ionization potentials of analyte gases. Consequently, different
sensor sets can be achieved. For example, amines, aromatic
compounds, and benzene are highly detectable at 9.5 eV.
Disease and other anomaly marker gases often found in the
breath, such as acetone, ammonia, and ethanol, are detectable at 9.5 eV as well as 10.6 eV. Other, more complex,
marker gases such as acetylene, formaldehyde, and methanol can be detected at 10.6 eV and 11.7 eV. Typically, the
PID devices are fairly responsive, on the order of a few
seconds, and do well with moderately low concentrations
(e.g., 0.1 ppm isobutylene).
One of the nice things about PIDs is that they can be
made very small in size, as shown in Fig. 3, which shows
the Rae Systems, Inc. ToxiRae personal gas monitor
(www.raesystems.com). Depending on the UV source,
CO, NO, SO2, or NO2 can be read. It works with rechargeable batteries.
Thermal Conductivity Detectors (TCD) are used to
detect and quantify gases that have large variations in
thermal conductivity. Gases that are discriminated well
are sulfur dioxide and chlorine, which have roughly onethird the conductivity of air to helium and hydrogen, which
have six and seven times the conductivity of air. As heat
transfer depends on three mechanisms, radiation, convection, and conduction, the actual TCD sensor itself must be
designed in such a way that conduction dominates, which
implies a very slow, constant, moving flow to minimize or
stabilize convection effects and a radiation-shielded enclosure. Most arrangements use two identically heated coils of
wire comprising two legs of a Wheatstone bridge, one coil in
a reference gas tube and the other in the sample tube.
When the thermal conductivity of the sample increases, the
sample coil is cooled more than the reference, and its
resistance changes (usually decreases), thereby generating
a voltage difference across the bridge. TCDs are usually
used in high concentration applications, as they do not
have the sensitivity of other techniques. TCDs do very well
when mounted at the exit of a GC where the separated gas
analytes are expected to have large variations in thermal
conductivity.
Gas chromatographs have come a long way over the last
decade as far as size and cost are concerned. Although
laboratory-grade devices such as the HP stand-alone system shown in Fig. 4 still are fairly common, portability is
being stressed in order to get the almost incomparable
Figure 4. An HP/Agilent laboratory-grade gas chromatograph.
detectibilty of the GC to where the real gas problems exist,

such as in the emergency or operating rooms, in the field,
and at sites where toxins and suspected hazardous gases
may be present. Bringing the GC to the patient or taking
data without having subjects come into the lab has
spawned systems such as Mensannas VOC (volatile
organic compound) GC system that uses a PID (Fig. 5)
to check trace gases in the breath, and HPs has introduced
a briefcase-sized micro-GC (Fig. 6). Lawrence Livermore
National Laboratory has taken the recent developments in
micromachining, MEMS (micro-electromechanical systems), and microfluidics and developed a real micro-GC.
Researchers at LLNL (4) have micro-machined a very long
capillary on a silicon chip, which serves as the separating
column.
Figure 5. Mensanna portable GC for measuring breath VOCs.
327
Table 1. Infrared Absorption Bands for Gases of Medical

Interest
Figure 6. Agilent (formerly HP) Micro-GC.
Figure 7 shows the implementation of this device that is

reported to have a response time of less than 2 min.
INFRARED/OPTICAL SPECTROSCOPY
Gases absorb light or photon energy at different wavelengths depending on the complexity of their molecular
structure. When a molecule is subjected to light energy of a
particular frequency, the atoms involved in the atomic
bonds will vibrate at the light frequency. If the frequency
matches their resonant frequency, or a harmonic, they will
resonate, thereby becoming highly absorbent as more of
the light energy is used to feed the motion of the resonating
molecules. The more complex a molecule, the greater
number of different atomic bonds it will have and,
consequently, the more absorption frequencies it will have.
Table 1 provides some guidelines for absorption for different molecules.
Most infrared analyzers measure concentrations of
volatile fluorocarbon halogenated anesthetic agents, carbon dioxide, and nitrous oxide using nondispersive infrared
(NDIR) absorption technology. The transduction means
may differ. Most use an electronic IR energy detector of
one sort or another, such as a bolometer, solid-state photon
detectors, and thermopiles; however, one monitor uses
Figure 7. LLNL MEMS-based micro-GC.
Wavelength
(microns)
Elements of Atomic
Bonds
2.74 mm
4.35 mm
5.266.6 mm
7.712.5 mm
X-H (X C, N, O, S)
C-X (X C, N, O)
C-X (X C, N, O)
C-X (X C, N, O)
Typical
Gases
H2O, CH4,
CO2, N2O
fluorocarbons
fluorocarbons
another IR detection principle, photoacoustic spectroscopy,

based on the level of sound produced when an enclosed gas
is exposed to pulsed/modulated IR energy.
Infrared analyzers have been used for many years to
identify and assay compounds for research applications.
More recently, they have been adapted for respiratory
monitoring of CO2, N2O, and halogenated anesthetic
agents.
Dual-chamber NDIR spectrometers pass IR energy from
an incandescent filament through the sample chamber and
an identical geometry but air-filled reference chamber.
Each gas absorbs light at several wavelengths, but only
a single absorption wavelength is selected for each gas to
determine the gas concentration. The light is filtered after
it passes through the chambers, and only that wavelength
selected for each gas is transmitted to the detector. The
light absorption in the analysis chamber is proportional to
the partial pressure (concentration) of the gas. Most manufacturers use a wavelength range around 3.3 mm, the
peak wavelength at which the hydrogen-carbon bond
absorbs light, to detect halogenated anesthetic hydrocarbons (halothane, enflurane, isoflurane, etc.).
In one monitor that identifies and quantifies halogenated anesthetic agents, the analyzer is a single-channel,
four-wavelength IR filter photometer. Each of four filters
(one for each anesthetic agent and one to provide a baseline
for comparison) transmits a specific wavelength of IR
energy. Each gas absorbs differently in the selected wavelength bands so that the four measurements produce a
unique signature for each gas. In another monitor, potent
anesthetic agents are assessed by determining their
absorption at only three wavelengths of light. Normally,
only one agent is present so this process reduces totagent
ID. However, the use of cocktails, mixtures of agents,
usually to reduce undesired side effects of one or another
agent, require very special monitoring because of the possibility of accidental overdosing.
The
Datex-Ohmeda
Capnomac
(www.us.datexohmeda.com), a multigas anesthetic agent analyzer, is
based on the absorption of infrared radiation. This unit
accurately analyzes breath-to-breath changes in concentrations of CO2, NO2, and N2O and anesthetic vapors. It is
accurate with CO2 for up to 60 breaths/min, and 30 breaths/
min for O2 (using a slower paramagnetic sensor), but N2O
and anesthetic vapors show a decrease in accuracy at
frequencies higher than 20 breaths/min. The use of narrow
wave-band filters to increase specificity for CO2 and N2O
makes the identification of the anesthetic vapors, which
are measured in the same wave band more difficult. It is
interesting to note that IR spectroscopy can also be used on
328
liquids, as exemplified by the Inov 3100 near-infrared

spectroscopy monitor that has been offered as a monitor
for intracerebral oxygenation during anesthesia and surgery. Studies with this monitor indicate that it needs a
wide optode separation and the measurements are more
likely those of the external carotid flow rather than the
divided internal carotid circulation (5).
A subset of NDIR is photoacoustic spectroscopy, which
measures the energy produced when a gas expands by
absorption of IR radiation, which is modulated at acoustic
frequencies. A rotating disk with multiple concentric
slotted sections between the IR source and the measurement chamber may be used tmodulate the light energy.
The acoustic pressure fluctuations created occur with a
frequency between 20 and 20,000 Hz, producing sound that
is detected with a microphone and converted totan electrical signal. Each gas (anesthetic agent, CO2, N2O) exhibits this photoacoustic effect most strongly at a different
wavelength. This method cannot distinguish which halogenated agent is present, however. The microphone detects
the pulsating pressures from all four gases simultaneously
and produces a four-component magnetic signal. A monitor
using IR photoacoustic technology has been developed that
can quantify all commonly respired/anesthetic gases
except N2 and water vapor. Similarly, a microphone detects
the pulsating pressure changes in a paramagnetic oxygen
sensor (magnetoacoustics).
The Bruel & Kjaer Multigas Monitor 1304 (6) measurements use photoacoustic spectroscopy and also incorporate
a pulse oximeter. It has some advantages over the Datex
Ohmeda Capnomac because it uses the same single microphone for detection of all gases, displaying gas concentration in real-time.
With the development of both fixed frequency and tunable solid-state lasers, a revolution in IR spectroscopy has
occured with new technical approaches appearing every
year. Tuned diode laser spectroscopy, and Laser-induced
Photo Acoustic Spectroscopy (a DARPA initiative) are
developments that bear close watching as they mature.
The ability to produce IR energy at extremely
narrow bandwidths allows discrimination of very closely
related gas species such as CO2 and N2O. In addition, most
of the volatile anesthetic agents such as halothane, desflurane, isoflurane, and sevoflurane can also be thus distinguished.
An advantage of NDIR is that the sensing mechanism
does not interfere or contact the sample, thus minimal
chance exists that the sample would be affected. The costs
of these devices have continued to decrease, with numerous
companies competing to keep the prices low and attractive.
Disadvantages are the susceptibility to dirt and dust in the
optical path and cross-sensitivities to interfering gases.
Companies marketing anesthesia and respiratory
mechanics monitors are involved in either development
or promotion of NDIR. Figure 8 shows a Datex-Ohmeda
Capnomac Ultima that uses NDIR for CO2, N2O and
anesthetic analysis, and agent identification. The top
waveform is the plethysmograph, the next down is the
O2 (measured with a fast paramagnetic sensor), and the
bottom waveform is the capnographic CO2 waveform. As an
added capability beyond gas analysis, to the right of the
Figure 8. A Datex-Ohmeda Capomac Ultima multiple gas

analyzer.
capnogram is the pressure-volume loop that is used to

assess the lung compliance.
Another limitation of NDIR is its relatively low sensitivity due, for the most part, to the short path length over
which the IR energy is absorbed. The short path length and
small chamber size is dictated by the need for fast response
in order to be able to monitor human physiological and
respiratory response.
Breath rates are normally about 10 breaths per min
(bpm) but, under acute hyperventilation conditions, can
reach 100 bpm and higher. Also, neonates and small animals naturally exhibit high breathing rates, which
requires a sensor with a millisecond response in order to
be able to detect breath-by-breath variations. However, in
cases where a fast response is not the driving factor, the
sampling chamber may be lengthened or the path length
increased.
Notwithstanding this limitation, the recent invention
of Cavity Ring-Down Spectroscopy (CRDS) by Princeton
chemist Kevin Lehmann (7,8) is based on absorption of
laser energy over huge path lengths but is extremely fast.
By bouncing laser energy between near-perfect mirrors in
a sample or test chamber, the light can pass through the
gas of interest multiple times, often for total distances of
up to 100 km, which creates the opportunity to detect
miniscule trace amounts of the gas species of interest.
The time it takes for the light energy to get attenuated
to zero provides a measure of the amount of the gas species
present. The shorter the Ring-Down time, the more of the
gas is present. These times are, however, on the order of
only milliseconds. In fact, Tiger Optics of Warrington, PA,
the company that has licensed Dr. Lehmanns technological development, claims that trace gases can be detected
in the hundreds of parts per trillion. The LaserTrace
multi-point, multi-species, multi-gas analyzer (shown in
Fig. 9) is capable of detecting many species such as H2O,
O2, CH4, H2, CO, NH3, H2S, and HF. The O2 module
measures down to 200 parts-per-trillion (ppt), in milliseconds, noninvasively and can readily be adapted to
respiratory breath measurements. Methane can be
detected at the parts per billion level.
Figure 9. Tiger Optics Cavity Ring-Down Spectrometer.
Although of interest in the detection and quantification

of oxygen, spectroscopy has not been widely considered for
this application. However, an oxygen absorption line exist
in the center of the visible spectrum at 760 nm. Often
neglected because of the problems associated with the
narrowness of the absorption band (0.01 nm versus
100 nm for CO2 in the IR) as well as with spurious interference from visible light, it is nonetheless an opportunity
because no interference exists from any other gases of
medical interest. The development of low cost, very narrow-band lasers has resulted in the successful introduction
of Laser-based Absorption Spectroscopy from Oxigraf
(www.oxigraf.com). With 100 ms response and 0.02%
resolution traces, such as that shown in Fig. 10, are possible. Cost is relatively low in comparison with other
technical approaches with similar capabilities.
LUMINESCENCE/FLUORESCENCE
Gas sensors that use luminescence or fluorescence basically take advantage of the phenomenon of excitation of a
molecule and the subsequent emission of radiation. Photoluminescence implies optical excitation and re-emission of
light at a different, lower frequency. Chemiluminescence
implies the emission of light energy as a result of a chemical reaction. In both cases, the emitted light is a function
of the presence of the gas species of interest and is detected
by optical means. Most industrial sensors use photomultiplier tubes to detect the light, but the needs of the medical
community are being met with more compact fiber-optic
systems and solid-state photodetectors.
Fluorescence quenching is a subset of luminescencebased sensors, the difference being that the presence of
the analyte, rather than stimulating emission of light,
329
actually diminishes the light output. Fundamentally,

fluorescence occurs when incoming light excites an electron
in a fluorescent molecule to a higher energy state, and, in
turn, when the electron returns to its stable state, it
releases energy in the form of light.
Two important characteristics of fluorescence are that
the light necessary to excite a fluorescent molecule has a
shorter wavelength than that of the fluorescent emission,
and that the fluorescence of a particular molecule may be
suppressed (quenched) or enhanced (dequenched) by the
presence of one or more specific molecules. Consequently,
the presence of such other molecules (called analytes) may
be detected.
A few companies exist that use one form or another of
luminescence sensing, in particular, of oxygen in the
medical arena, although the sensors are ubiquitous for
other gases. For example, Ocean Optics (www.oceanoptics.com) FOXY Fiber Optic Oxygen Sensors use the fluorescence of a ruthenium complex in a sol-gel to measure the
partial pressure of oxygen. First, a pulsed blue LED sends
light, at 475 nm, to and through an optical fiber, which
carries the light to the probe. The distal end of the probe tip
consists of a thin layer of a hydrophobic sol-gel material. A
ruthenium complex is trapped in the sol-gel matrix, effectively immobilized and protected from water. The light
from the LED excites the ruthenium complex at the probe
tip and the excited ruthenium complex fluoresces, emitting energy at 600 nm. When the excited ruthenium
complex encounters an oxygen molecule, the excess energy
is transferred to the oxygen molecule in a nonradiative
transfer, decreasing or quenching the fluorescence signal.
The degree of quenching is a function of the level of oxygen
concentration pressure in the film, which is in dynamic
equilibrium with oxygen in the sample. Oxygen as a triplet
molecule is able to efficiently quench the fluorescence and
phosphorescence of certain luminophores. This effect is
called dynamic fluorescence quenching. When an oxygen
molecule collides with a fluorophore in its excited state, a
nonradiative transfer of energy occurs. The degree of
fluorescence quenching relates to the frequency of collisions and, therefore, to the concentration, pressure, and
temperature of the oxygen-containing media. The energy
is collected by the probe and carried through an optical
fiber to a spectrometer where an analog-to-digital (A/D)
converter converts the data to digital data for use with a
PC.
MASS SPECTROSCOPY
Figure 10. Oxigraf Fast Oxygen Sensor trace of respired O2.
Mass spectroscopy provides, what many consider, the best

accuracy and reliability of all of the gas analyzing/assaying
schemes. The basic concept is to assay the analyte by
reducing it into ionized component molecules and separating them according to their mass-to-charge ratio. By this
technique, the constituents of the sample gas are ionized.
The resulting ions are accelerated through an electrostatic
field and then passed through a deflecting magnetic field.
The lighter ions will deflect more than the heavier ions.
Detecting the displacement and counting the ions can
achieve the assay of the gas sample.
330
Ion detectors usually comprise an electric circuit where

the impinging ions generate a current, which can be
measured with a conventional circuit. The more current,
the more ions are impinging. In practice, the ionization
must be conducted in a high vacuum of the order of 106
torr. The gas is ionized with a heated filament, or by other
means as have been discussed before.
A number of different mass spectroscopic configurations have been developed over the years. Time-of-flight
(TOF) systems differ from the magnetically deflected
devices in that the ions are free to drift across a neutrally
charged evacuated flight chamber after having been accelerated electrostatically by a series of gratings that separate the ions. The time it takes for the ions to travel across
the chamber is a function of their mass. An output not
unlike that of a GC is developed. Quadrupole mass spectrometers focus the ions through an aperture onto a quadrupole filter. The ion-trap mass spectrometer traps ions in a
small volume using three electrodes. An advantage of the
ion-trap mass spectrometer over other mass spectrometers is that it has a significantly increased signal-tonoise ratio because it is able to accumulate ions in the trap.
It also does not require the same kind of large dimensions
that the TOF and magnetically deflected devices need, so,
as a consequence, it can be made in a fairly compact size.
Finally, the Fourier-transform mass spectrometer takes
advantage of an ion-cyclotron resonance to select and
detect ions. Single-focusing analyzers use a circular beam
path of 1808, 908, or 608. The various forces influencing the
particle separate ions with different mass-to-charge
ratios. Double-focusing analyzers have an electrostatic
analyzer added to separate particles with difference in
kinetic energies.
A particular advantage of the mass spectrometer is that
it can operate with an extremely small gas sample and can
detect minute quantities. Response was long compared
with most continuous sensors, but with the development
of high speed microprocessors, analysis times have steadily
decreased to where, today, it is not unusual to have assays
in less than one minute. With the development of MEMS
TOF devices, the time-of-flight is measured in microseconds.
Mass spectrometers have always tended to be bulky and
expensive and, thus, rarely used on a single patient basis.
Multiplexing up to 30 patients using a complex valving
switching system has been shown to be feasible and has
made the system much more cost-effective. Figure 11
shows a conventional ThermoElectron laboratory-grade
mass spectrometer setup.
The move to miniature mass spectrometers has been
rapid over the last decade, from a suitcase-sized miniature
TOF mass spectrometer, developed at Johns Hopkins
Applied Physics laboratory (Fig. 12) (9), to a micro-electromechanical system (MEMS) device smaller than a penny,
developed in Korea at the MicroSystems Lab of Ajou University (Fig. 13) (10).
Mass spectroscopy is often used as the detector in
combination with gas chromatography to enhance the
sensitivity down to ppb, because in a GC, detection limits
the capability.
Figure 11. A ThermoElectron laboratory-grade quadrupole mass

spectrometer.
Figure 12. JHU Teeny mass spectrometer.
NUCLEAR MAGNETIC RESONANCE

Nuclear Magnetic Resonance (NMR) is the process by
which a relatively low intensity radio-frequency (RF) signal at the resonant frequency of the species of gas of
interest interacts with the atoms in a gas and aligns them
momentarily, which requires some energy. When the RF
Figure 13. MEMS TOF mass spectrometers developed at Ajou

University in Korea.
signal is removed, the atoms release the energy that had

been stored in the alignment resonance, return to their
chaotic orientations, and re-emit an RF signal at the same
resonant frequency at which they were excited. Each
atomic bond has its own characteristic frequency so that
a spectroscopic analysis can be made by scanning through a
large number of frequencies. A variant of NMR, nuclear
quadrupole resonance (NQR) is used for detecting explosive vapors, which could be very useful in military medicine
where explosives in clothing during triage pose a major
hazard. The hydrogen bonds in TNT have a resonance at
760 kHz and the vapors of plastic explosives have resonances at low MHz frequencies. NMR is very attractive
because gas analysis can be performed without having to
physically take a sample of the analyte in question. As the
initiating and re-emitted RF signals can pass through most
nonferrous materials with little attenuation, gas, liquid,
and solid chemical species can be interrogated noninvasively. By summing the returns from a number of signals,
sensitivity to the low ppm can be achieved.
In dirty environments where optical or infrared devices
suffer significant degradation of performance, NMR is
particularly useful. Compared with chromatographic
approaches, NMR eliminates the need for solvents, columns, carrier gases, or separations. Also, NMR analysis
can be performed in real-time because typical atomic
relaxation times, the time it takes for the atomic spin axes
to return to their original orientations, is on the order of
milliseconds for many gases of interest.
PARAMAGNETISM
Many gases exhibit magnetic sensitivity, paramagnetism,
due to their polar nature, which means that they are
attracted to a magnetic field. For oxygen, its paramagnetic
sensitivity is due to its two outer electrons in unpaired
orbits. Most of the gases used in anesthesia are repelled by
a magnetic field (diamagnetism).
Paramagnetic sensors are typically used specifically for
measuring oxygen concentration. The high degree of sensitivity of oxygen (compared with other gases) to magnetic
forces reduces the cross-sensitivity to other gases of paramagnetic sensors. Sensors come in two variants, the older
balance type and the newer pressure type. The balance
types of sensors are relatively frail and have been replaced
by the pressure types. Nevertheless, some of these older
devices are still being used. The balance type of sensor uses
a mechanical approach that has a dried gas sample flowing
through a chamber in which a nitrogen-filled dumbbell is
balanced in a magnetic field. The paramagnetic force on the
oxygen in the sample puts a torque on the dumbbell. The
output can be read either as a spring-loaded displacement
or, in newer devices, electronically by measuring the current required to keep the dumbbell centered.
Most modern paramagnetic oxygen sensors consist of a
symmetrical, two-chambered cell with identical chambers
for the sample and reference gas (often air or nitrogen).
These cells are joined at an interface by a responsive
differential pressure transducer or microphone. Sample
and reference gases are pumped through these chambers
331
in which a strong, usually varying, magnetic field surrounding the region acts on the oxygen molecules and
generates a static pressure or a time-varying (acoustic)
difference between the two sides of the cell, causing the
transducer to produce a DC or AC voltage proportional to
the oxygen concentration. When the magnetic field is
modulated at acoustic frequencies, these devices may
sometimes be referred to as magnetoacoustic.
Paramagnetic oxygen analyzers are very accurate,
highly sensitive, and responsive, often with a step response
of 200 ms to 90% of maximum. However, they require
calibration, usually with pure nitrogen and oxygen. A
major drawback is that they are adversely affected by
water vapor and, consequently, require a water trap incorporated into their design. The frequency response makes
then useful for measurement of oxygen on a breath-bybreath basis. The Datex Ohmeda Capnomac Ultima that
was previously shown in Fig. 8 uses a paramagnetic oxygen
sensor, as do many of the other mainline medical gas
monitor manufacturers.
RADIOACTIVE IONIZATION
The ubiquitous smoke detector found in every house, hospital, and facility has spawned detectors for other gases
such as carbon monoxide, a very important marker gas in
medical diagnosis. Although usually used as devices that
are set to alarm when a preset level is detected, they are
also used as calibrated sensors. A very low level radioactive
alpha particle source (such as Americium-241) can ionize
certain gases so that, in the presence of an analyte, a circuit
can be completed and current caused to flow in the detector
circuit.
Ionization detectors detect the presence of invisible
particles (less than 0.01 micron in size) in the air. Inside
the detector, a small ionization chamber exists that contains an extremely small quantity of radioactive isotope.
Americium-241 emits alpha particles at a fairly constant
rate. The alpha particles, which travel at an extremely
high rate of speed, knock off an electron from the oxygen
and nitrogen molecules in the air passing through the
ionization chamber. The free electron (negative charge)
is then attracted to a positively charged plate, and the
positively charged oxygen or nitrogen is attracted to a
negatively charged plate, which creates a very small but
constant current between the plates of a detector circuit,
which in itself is a gas detection mechanism much in the
same way that the other ionization detectors operated.
However, when particles, such as soot particles, dust,
fumes, or steam, enter the ionization chamber, the current
is disrupted. If the current decreases too mid, an alarm is
triggered.
The disadvantage of these devices is clearly the health
hazard associated with the presence of the radioactive
material. However, because the detector contains only a
tiny amount of radioactive material, exposure is unlikely
with proper care in handling. Another disadvantage of
these sensitive detectors is the false-positive alarms that
can be triggered by spurious dust and other nontoxic
fumes. However, the big advantage is that ionization
detectors are very sensitive and, given that false alarms
332
are tolerable, should be considered in most alarm situations.

Another form of radioactive detector is the electron
capture detector, which uses a radioactive Beta emitter
(electrons) to ionize some of the carrier gas and produce a
current between a biased pair of electrodes. When organic
molecules that contain electronegative functional groups,
such as halogens, phosphorous, and nitro groups, pass by
the detector, they capture some of the electrons and reduce
the current measured between the electrodes.
RAMAN LASER SPECTROSCOPY
In the 1980s, Raman scattering was first heralded as an
improvement to mass spectrometry (11), although some
individuals had reservations (12). Although no longer
manufactured but still serviced, Ohmeda Rascal II
multi-gas analyzer uses a Raman scattering of laser light
to identify and quantify O2, N2, CO2, N2O, and volatile
anesthetic agents. It is stable and can monitor N2 directly
and CO2 accurately for a wide range of concentrations. One
of the acknowledged disadvantages is that a possibility of
some destruction of volatile anesthetic agent exists during
the analysis because the concentration of halothane does
appear to fall when recirculated and as much as 15% must
be added. Some concern exists over the reliability of the
hardware, software, and laser light source (13) that is
currently being addressed by others.
Raman scattering occurs when a gas sample is drawn
into an analyzing chamber and is exposed to a high intensity beam from an argon laser. The laser energy is absorbed
by the various molecules in the sample and are then excited
into unstable vibrational or rotational energy states, which
is the Raman scattering. The low intensity Raman scattered, or re-emitted, light signals are measured at right
angles to the laser beam, and the spectrum of Raman
scattering lines can be used to identify various types of
gas molecules. Spectral analysis allows identification of
known compounds by comparison with their Raman spectra. This technique is of similar accuracy to mass spectrometry.
SOLID-STATE SENSORS
At least four types of solid-state gas sensors exist: semiconductor metal oxide (SMO) sensors; chemically sensitive
field effect transistors (ChemFETs); galvanic oxide sensors; and piezoelectric or surface acoustic wave (SAW)
crystal sensors.
Semiconductor metal sensors are an outgrowth of the
development of semiconductor devices. Early in the development of transistors and integrated circuits, it was
observed that the characteristics would change in the
presence of different gases. Recalling that a transistor is
basically a voltage-controlled resistor, it was discovered
that the p-n junction resistance was being changed by
chemical reaction with the semiconductor materials (14).
Commercially available Taguchi Gas Sensors (TGS) tin
oxide sensors have found a niche as electronic noses.
Walmsley et al. (15) used arrays of TGS sensors to develop
patterns for ether and chloroform and other vapors of

medical interest.
Hydrocarbons were among the first gases to be detected,
and later, hydrogen sulfide was found to be detectable.
Since the first tin oxide sensors appeared in the late 1960s,
it has been found that by doping transition metal oxides,
such as tin and aluminum, with other oxides, that as many
as 150 different gases could be specifically detected (1) at
ppm levels. The heated oxide adsorbs the analyte and the
resistance change is a function of the concentration of the
analyte. The semiconducting material is bonded or painted
in a paste to a nonconducting substrate and mounted
between a pair of electrodes. The substrate is heated to
a temperature such that the gas being monitored reversibly changes the conductivity of the semiconducting metal
oxide material. When no analyte is present, the current
thinking is that oxygen molecules capture the free electrons in the semiconductor material when they are absorbing on the surface, thereby preventing the mobility of the
electron flow. Analyte molecules replace the oxygen,
thereby releasing the free electrons and, consequently,
reducing the SMO resistance between the electrodes.
The ChemFET is derived from a field effect transistor
where the normal gate metal has been replaced with a
catalytic metal or chemically sensitive alloy. The gaseous
analyte interacts with the gate metal and changes the FET
characteristics to include gain and resistance.
Solid-state galvanic cells are based on the semiconductor galvanic properties of certain oxides or hydrides.
The zirconium oxide galvanic cell oxygen sensor is probably
one of the most ubiquitous sensors in daily life. It is the
sensor mounted in every automobile catalytic converter to
measure its effectiveness. Zirconium oxide, when heated to
a temperature of about 700 8C, becomes an oxygen ion
conductor, so that, in the presence of a difference in partial
pressure on either side of a tube with metallized leads
coming off each side, a voltage potential (Nernst voltage) is
developed. These devices are commonly called fugacity
sensors. As the process is reversible, a voltage applied will
cause oxygen ions to flow. This process may also be applicable to the hydrogen ions in hydrides. An advantage of
these oxygen sensors over other types is that no consumable exists. Hence, life is long. However, the need for
heating tends to make these devices difficult to handle
and they, as well as SMOs, require significant power to
power the heating elements. However, because these sensors can be very small, they can have fast response times,
often less than 100 ms, which makes them suitable for use
for respiration monitoring. The electronics associated with
the detection circuits is simple and should be very reliable.
Piezoelectric sensors use the change in a crystal vibrational frequency or propagation of surface acoustic waves
to measure the concentration of a selected analyte. Most
often, an analyte sample is passed through a chamber
containing two piezoelectric crystals: a clean reference
crystal and a second crystal that has been coated with a
compound that specifically adsorbs specific analyte gases.
Organophillic coatings are used for hydrocarbons such as
anesthetic vapors. The resulting increase in mass changes
the coated crystals resonant frequency or the speed of
propagation in direct proportion to the concentration of
333
anesthetic gas in the sample. Using either some form of

beat frequency measurement or detection of the phase shift
between the two crystals, a detection circuit can generate a
signal that can be processed and displayed as a concentration.
EMERGING TECHNOLOGIESMEMS
MICROFLUIDICSNANOTECHNOLOGY
The development of a variety of microfluidic labs-on-a-chip
is leading the charge in the state-of-the-art in gas sensing.
It was noted in the gas chromatography section that microfluidic channels are being used as GC columns and in the
mass spectrometry section that microfluidics plays a major
role in the TOF passages and chambers. The ability to
miniaturize classic technologies has opened the door to
mass production as well as the ability to mix-and-match
sensors and technologies. The development of electronic
noses that can discriminate between thousands of different
chemicals and gases is driving the need to detect odors and
minute quantities of dangerous or toxic gases.
Patient safety in medicine continues to be a major
driver. MEMS (micro-electromechanical systems) and
nanotechnology have become the enabling technologies
for placing thousands of sensors in microdimensional
arrays that can be placed inside a capsule and swallowed
or implanted to monitor physiological and metabolic
processes. IR bolometers that can sense incredibly small
temperature differences as low as 0.02 8C are already a part
of todays inventory (Fig. 14), and in the future, nanotechnology elements that are merely one molecule thick and
dust particle-sized may provide IR spectroscopic capability
in implantable or inhaled micro-packages.
A new paradigm for gas analysis that has been enabled
by the development of microfluidics was originally suggested in the 1960s by Mapleson (16), who suggested
measuring a physical gas property as a way of inferring
binary gas mixture composition. This concept has been
extended and implemented for ternary and quaternary
gas mixtures with a microfluidic gas multiple gas property
analyzer (1720). Ternary mixtures are assayed with a
chip that measures viscosity with a microcapillary viscometer and density with a micro-orifice densitometer. An
early prototype microfluidic lab-on-a-chip showing the
microcapillaries is shown in Fig. 15. By measuring properties possessed in common by all gases, such as density,
viscosity, and specific heat, a single chip can be used to
Figure 14. 25 micron wide longwave IR microbolometer detector.
Figure 15. Microfluidic chip that measures the density and

viscosity of three-component respired gases from which the
constituent gas concentrations of oxygen, nitrogen, and carbon
dioxide are deduced.
analyze mixtures of any four gases. The concentrations of

the constituents can be determined by simultaneously
solving the equations that relate the mixture properties
to the concentrations, thereby determining the relative
concentrations of the mixture gases required to produce
the measured properties. The only limitation to such an
approach is that one must know what at least all but one of
the constituents are. Microfluidic property sensors such a
capillary viscometers, orifice densitometers, and speed-ofsound calorimeters can provide real-time simultaneous
assays of respiratory gases (O2, CO2, and N2), the simultaneity of which then enables the reduction to practice of a
variety of physiologic analyzers such as metabolic rate,
cardiac output, and cardio-pulmonary function that had
been postulated back in the 1960s (21) but never practically
implemented.
Advances in optics and optical fiber technology, greater
expansion of solid-state sensing, and the revolutionary
aspects of nanotechnology will provide gas analysis and
sensing with new capabilities, provided that the lessons
learned from the past are appropriately heeded.
OTHER CONSIDERATIONS IN GAS ANALYSIS
Most continuous gas analysis devices are side-stream
monitors that acquire gas samples from a breathing circuit
through long, narrow-diameter tubing lines. These lines
may incorporate moisture removal to allow moisture to
pass through the sampling line and into the atmosphere, as
is the case with Nafion tubing. A water trap or filter may
also be used to remove condensation from the sample in
order to reduce water vapor before the gas sample reaches
the analysis chamber. Gas samples are aspirated into the
monitor at either an adjustable or a fixed-flow rate, typically from 50 to 250 ml/min. Lower rates minimize the
amount of gas removed from the breathing circuit and,
therefore, from the patients tidal volume; however, lower
sampling flow rates increase the response time and typically reduce the accuracy of conventional measurements.
334
Gas monitors have to eliminate the exhaust gas through a

scavenging system or back to the patients breathing circuit.
DISPLAYS, ALARMS, CALIBRATION, AND CONTROLS
Many gas monitors provide a graphic display of breath-bybreath concentrations and a hardcopy of trends of gas
concentrations from the beginning to the end of an event
(e.g., anesthesia delivery during an operation). The user
typically calibrates or verifies calibration of the sensors
with a standard gas mixture from an integral or external
gas cylinder. Gas monitors are usually microprocessorcontrolled and have user-adjustable alarms that typically
include factory-preset default alarms or alarm-set programs for both high and low concentrations of the gases
measured. Some monitors also have alarms for system
malfunctions, such as disconnection from a gas source,
and leaks can often be identified from trending of O2
and CO2. Occlusion, apnea, or inadvertent rebreathing
can also be identified. Most monitors typically have a
method for temporarily, but not indefinitely, silencing
audible alarms for low O2, high N2O, and high agent,
whereas other, less critical audible alarms can be permanently silenced. Most monitors typically display real-time
waveforms and long-term trends. They have integral display capability and are also commonly equipped with output jacks to interface with computerized record-keeping
systems or with additional analog or digital display units
such as chart recorders and printers.
PATIENT SAFETY
A review of the background and significance of medical gas
sensors and monitors would be incomplete without an
expression of the context that patient safety has had on
the impetus for recent gains in technology and the need for
additional improvements. Clearly the intrinsic dangers in
the conduct of anesthesia have been long understood. It
became evident in the early 1980s that patient safety and
reduction to risk was possible if respiratory and anesthetic
gas monitoring was routinely available and used. As a
result of improved and increased availability of medical
gas monitoring technology and professional activity lead by
the Anesthesia Patient Safety Foundation (APSF) with the
support of the American Society of Anesthesiologists
(ASA), a standard for monitoring has been adopted and
is routinely used in todays clinical practice. This standard
requires assessment of the patients ventilation and oxygenation in addition to circulation and temperature. The
use of such monitors has resulted in a significant decrease
in the risk of anesthesia-related deaths and morbidity in
the ICU and other critical care situations.
CONCLUSION
Medical gas monitoring has been so successful in improving patient safety and reducing patient risk that medical
malpractice liability insurance companies have lowered
their risk liabilities and premiums to anesthesiologists

who guarantee the routine implementation of these standards whenever possible (22). The argument for providing
additional patient safety will continue to be a powerful
incentive to improve and enhance the methods and techniques to provide increased knowledge of the monitoring of
respiratory and anesthetic gases.
The availability of gas sensors and monitors is a boon to
the medical profession from both a clinical as well as a
research point of view. In addition to patient safety, new
diagnostic capabilities are emerging every year. In
research, new gas-sensing capabilities are enhancing the
discovery of markers for all kinds of disease states and
metabolic functions.
Looking to the future, MEMS, microfluidics, and nanotechnology will provide growth in our understanding of
physiologic processes at levels of detail never before conceived of, from inside the body as well as supplanting
todays conventional techniques.
BIBLIOGRAPHY
1. Chou J, Hazardous Gas Monitors, A Practical Guide to Selection, Operation and Applicationsy. New York: McGraw-Hill;
2000.
2. ECRI. Multiple medical gas monitors, respired/anesthetic.
Product Comparison System, ECRI Product Code 17445,
August 1993.
3. Datex Ohmeda Smart Vent 7900. Product Description
AN2842-B/7 01 1999 Datex-Ohmeda, Inc.
4. Yu CM, Sheem SK. (1995). Miniature gas chromatography
sensor. Lawrence Livermore National Lab. www.llnl.gov/sensor_technology/STR72.html.
5. Harris D, Bailey S. Near infrared spectroscopy in adults.
Anaesthesia 1993;48:694696.
6. McPeak HB, Palayiwa E, Robinson GC, Sykes MK. An evaluation of the Bruel and Kjaer monitor 1304. Anaesthesia
1992; 47(1):4147.
7. Lehmann KK, Rabinowitz P. High-finesse optical resonator
for cavity ring-down spectroscopy based upon Brewsters
angle prism retroreflectors. U.S. Patent 5,973,864, October
26,
1999.
8. Lehmann KK, Romanini D. The superposition principle and
cavity ring-down spectroscopy. J Chem Phys 1996;105(23):
1026310277.
9. Ecelberger SA, Cornish TJ, Bryden W. The improved teenyTOF mass spectrometer for chemical and biological sensing,
3rd Harsh-Environment Mass Spectrometry Workshop,
March 2528, 2002 Pasadena, CA.
10. Yoon HY, Kim JH, Choi ES, Yang SS, Jung KW. Fabrication
of a novel micro time-of-flight mass spectrometer. Sens
Actuators A 2002;9798:441447.
11. Westenskow DR, Smith KW, Coleman DL, et al. Clinical
evaluation of a Raman scattering multiple gas analyzer.
Anesthesiology 1989;70:350355.
12. Severinghaus JW, Ozanne GM. Multi-operating room monitoring with one mass spectrometer. Acta Anaesthesiol Scan
1987; (Suppl) 70:186187.
13. Lockwood G, Landon MJ, Chakrabarti MK, Whitwam JG.
The Ohmeda Rascal II. Anaesthesia 1994;49:4453.
14. Kress-Rogers E, ed. Handbook of Biosensors and Electronic
NosesMedicine, Food and the Environment. Boca Raton,
FL: CRC Press; 1996.

15. Walmsley AD, Haswell SJ, Metcalfe E. Methodology for the
selection of suitable sensors for incorporation into a gas
sensor array. Analytica Chimica Acta 1991;242:31.
16. Mapleson WW. Physical methods of gas analysis. Brit J
Anaesth 1962;34:631.
17. Drzewiecki TM. Fluidic multiple medical gas monitor. NIH
BioEngineering Symposium, Bethesda, MD, February 1998.
18. Drzewiecki TM, Polcha M, Koser M, Calkins J. A novel
inexpensive respiratory and anesthetic gas monitor. Society
for Technology in Anesthesia 1999 Annual Meeting. San
Diego, CA: January 1999.
19. Drzewiecki TM, Calkins J. Real time, simultaneous analysis
of multiple gas mixtures using microfluidic technology. Proc
Instrument Society of America AD 2000 Symposium, Charleston, WV: April 2000.
20. Drzewiecki TM. Method and apparatus for real time gas
analysis. U.S. Patent 6,076,392, June 2000.
21. Kim TS, Rahn H, Farhi LE. Estimation of true venous and
arterial PCO2 by gas analysis of a single breath. J Appl
Physiol 1966;21(4):13381344.
22. Swedlow DB. Respiratory gas monitoring. In: Saidman L,
Smith N, eds. Monitoring in Anesthesia. 3rd ed. Boston, MA:
Butterworth-Heinemann; 1993. pp 2750.
See also BLOOD
GAS MEASUREMENTS; RESPIRATORY MECHANICS AND GAS
EXCHANGE.
MEDICAL PHOTOGRAPHY. See PHOTOGRAPHY,

MEDICAL.

COLIN ORTON
Harper Hospital and Wayne
State University
Detroit, Michigan
INTRODUCTION
Medical physicists are responsible for the application of
physics to the diagnosis and treatment of disease and other
disabilities although, in some countries, the treatment of
patients with disabilities is a separate field, often referred
to as biomedical engineering or words to that effect. Here,
we restrict ourselves to applications in the diagnosis and
treatment of disease.
The major applications in diagnosis are the use of
X-rays for imaging (diagnostic radiology, including
computerized tomography (CT), etc.); radioactive isotopes
for imaging and uptake measurements [nuclear medicine,
single photon emission computed tomography (SPECT),
positron emission tomography (PET), etc.]; magnetic resonance magnetic resonance imaging (MRI) and spectroscopy (MRS); ultrasound (ultrasonography).
Applications of physics to the treatment of disease
include the following: external beams of X-rays, gammarays, or electrons for the treatment of cancer radiation
oncology, including stereotactic radiosurgery, intensity
335
modulated radiation therapy (IMRT), total body irradiation (TBI), etc.; external beams of heavy particles
for the treatment of cancer (neutrons, protons, heavy
ions); internal radioisotope treatments for cancer
(brachytherapy, systemic radiotherapy, and radioimmunotherapy) and other problems (intravascular brachytherapy, hyperthyroidism, etc.); hyperthermia for the
treatment of cancer.
Other topics of major interest for medical physicists
include various medical applications of light and lasers,
radiation protection, radiation measurements, radiation
biology, and the applications of computers to all of the
above.
Throughout the world there are medical physics organizations that represent medical physicists and provide
information to help them in their profession. Most of
these are national associations, with about 70 of these
represented by the International Organization for Medical
Physics (IOMP). In terms of publications, by far the two
most prolific organizations are the Institute of Physics
and Engineering in Medicine (IPEM) in the United
Kingdom, and the American Association of Physicists in
Medicine (AAPM) in North America. These two organizations between them have published hundreds of
reports, monographs, and meeting proceedings that are
used as reference materials throughout the world. From
the IPEM many of these are published by the Institute of
Physics Publishing (IOPP) in Bristol, UK, and from the
AAPM many are published by either the American Institute of Physics (AIP) or Medical Physics Publishing
(Madison, WI).
JOURNALS
Several national and international organizations and
independent publishers publish journals used by medical
physicists. Some of these journals are used extensively
for medical physics papers in which at least 25% of the
manuscripts are medical physics articles. These are categorized as Primary journals below. Others contain some
medical physics articles (<25%) and are categorized as
Secondary.
PRIMARY MEDICAL PHYSICS JOURNALS
Australasian Physical & Engineering Sciences in Medicine, Australasian College of Physical Scientists and
Engineers in Medicine and the College of Biomedical
Engineers.
Journal of Applied Clinical Medical Physics, American
College of Medical Physics (http://www.jacmp.org).
Journal of Medical Physics, Association of Medical Physicists of India.
Medical Dosimetry, American Association of Medical
Dosimetrists (Elsevier).
Medical Physics, American Association of Physicists in
Medicine (AIP).
Physica Medica, Istituti Editoriali e Poligrafici Internazionali Casella Postale n.1, Succursale n.8, 56123
Pisa, Italy (http://www.iepi.it).
336
Physics in Medicine and Biology, Institute of Physics

and Engineering in Medicine (IOPP).
Polish Journal of Medical Physics and Engineering,
Polish Society of Medical Physics.
sterZeitschrift fu r Medizinische Physik, Deutschen, O
reichischen und Schweizerischen Gesellschaft fu r
Medizinische Physik (Elsevier).
SECONDARY MEDICAL PHYSICS JOURNALS
Acta Radiologica, Scandinavian Society of Radiology
(Taylor & Francis).
American Journal of Roentgenology, American Roentgen Ray Society.
Applied Radiation and Isotopes (Elsevier).
Australasian Radiology, Royal Australian and New
Zealand College of Radiologists (Blackwell).
Biomedical Imaging and Interventional Journal (University of Malaya: http://www.biij.org).
Biomedical Instrumentation & Technology, Association
for the Advancement of Medical Instrumentation.
Brachytherapy, American Brachytherapy Society
(Elsevier).
British Journal of Radiology, British Institute of
Radiology.
Canadian Association of Radiologists Journal, Canadian Association of Radiologists.
Cancer Radiothe rapie, Socie te Francaise de Radiothe rapie Oncologique (Elsevier).
Cardiovascular and Interventional Radiology, Cardiovascular and Interventional Radiological Society of
Europe, Japanese Society of Angiography and Interventional Radiology, and British Society of Interventional Radiology (Springer).
Cardiovascular Radiation Medicine (Elsevier).
Clinical Imaging (Elsevier).
Clinical Radiology, Royal College of Radiologists
(Elsevier).
Critical Reviews in Computed Tomography (Taylor &
Francis).
Computerized Medical Imaging and Graphics, Computerized Medical Imaging Society (Elsevier).
Computers in Biology and Medicine (Elsevier).
Current Problems in Diagnostic Radiology (Mosby).
European Journal of Nuclear Medicine and Molecular
Imaging, European Association of Nuclear Medicine
(Springer).
European Journal of Radiology (Elsevier).
European Journal of Ultrasound, European Federation
of Societies for Ultrasound in Medicine and Biology
(Elsevier).
European Radiology, European Congress of Radiology
(Springer).
Health Physics, Health Physics Society (Lippincott
Williams & Wilkins).
IEEE Transactions on Medical Imaging, IEEE.

International Journal of Radiation Biology (Taylor &
Francis).
International Journal of Radiation Oncology, Biology,
Physics, American Society of Therapeutic Radiology
and Oncology (Elsevier).
Investigative Radiology (Lippincott Williams & Wilkins).
Journal of Biomedical Optics, International Society for
Optical Engineering (SPIE).
Journal of Cardiovascular Magnetic Resonance, Society
for Cardiovascular Magnetic Resonance (Taylor &
Francis).
Journal of Clinical Ultrasound (Wiley).
Journal of Computer Assisted Tomography (Lippincott
Journal of Diagnostic Radiography and Imaging, Royal
Society of Medicine.
Journal of Digital Imaging, Society for Computer Applications in Radiology (Springer).
Journal of Electronic Imaging, International Society for
Optical Engineering (SPIE).
Journal of Labelled Compounds and Radiopharmaceuticals (Wiley).
Journal of Magnetic Resonance Imaging, International
Society for Magnetic Resonance Medicine (Wiley)
Journal of Neuroimaging, American Society of Neuroimaging (Sage Publications).
Journal of Nuclear Cardiology, American Society of
Nuclear Cardiology (Elsevier).
Journal of Nuclear Medicine, Society of Nuclear
Medicine.
Journal of Nuclear Medicine Technology, Society of
Nuclear Medicine.
Journal of Radiological Protection, Society for Radiological Protection (IOPP, Bristol, U.K.).
Journal of the Acoustical Society of America, Acoustical
Society of America (American Institute of Physics,
New York).
Journal of the American Society of Echocardiography, American Society of Echocardiography
(Mosby).
Journal of Thoracic Imaging, Society of Thoracic
Radiology (Lippincott Williams & Wilkins).
Journal of Ultrasound in Medicine, American Institute
of Ultrasound in Medicine.
Journal of Vascular and Interventional Radiology,
Society of Interventional Radiology (Lippincott
Journal of X-Ray Science and Technology (IOS Press,
Amsterdam).
Lasers in Medical Science (Springer).
Lasers in Surgery and Medicine, American Society for
Laser Medicine and Surgery (Wiley).
Medical Engineering & Physics (Elsevier).
Magnetic Resonance Imaging (Elsevier).
Magnetic Resonance in Medicine, International Society

for Magnetic Resonance in Medicine (Wiley).
Medical Engineering & Physics, Institute of Physics and
Engineering in Medicine (Elsevier).
Medical Image Analysis (Elsevier).
Molecular Imaging and Biology, Academy of Molecular
Imaging (Springer).
Neuroradiology, European Society of Neuroradiology
(Springer).
NMR in Biomedicine (Wiley).
Nuclear Medicine and Biology, Society of Radiopharmaceutical Sciences (Elsevier).
Pediatric Radiology, European Society of Pediatric
Radiology, Society for Pediatric Radiology, Asian
and Oceanic Society for Pediatric Radiology
(Springer).
Photomedicine and Laser Surgery, World Association
for Laser Therapy (Mary Ann Liebert, Inc.).
Physiological Measurement, Institute of Physics and
Engineering in Medicine (IOPP).
Progress in Nuclear Magnetic Resonance Spectroscopy
(Elsevier).
Radiation Measurements (Elsevier).
Radiation Physics and Chemistry (Elsevier).
Radiation Research, Radiation Research Society.
Radiographics,
Radiological
Society
of
North
America.
Radiography, College of Radiographers (Elsevier).
Radiology, Radiological Society of North America.
Radiotherapy and Oncology, European Society
for
Therapeutic
Radiology
and
Oncology
(Elsevier).
Seminars in Interventional Radiology (Thieme).
Seminars in Nuclear Medicine (Elsevier).
Seminars in Radiation Oncology (Saunders).
Seminars in Roentgenology (Elsevier).
Seminars in Ultrasound, CT and MRI (Elsevier).
Techniques in Vascular and Interventional Radiology
(Elsevier).
Topics in Magnetic Resonance Imaging (Lippincott
Ultrasound in Medicine & Biology, World Federation for
Ultrasound in Medicine and Biology (Elsevier).
Ultrasound in Obstetrics and Gynecology (Wiley).
Year Book of Diagnostic Radiology (Elsevier).
Year Book of Nuclear Medicine (Elsevier).
BOOKS AND REPORTS

As with journals, books and reports are published by both
medical physics organizations, especially the AAPM and
the IPEM, and independent publishers. Most of the books
and reports in the following lists can be purchased directly
337
from the publishers or, alternatively, through bookstores

using the ISBN number provided.
MEDICAL AND RADIOLOGICAL PHYSICS
General
A Century of X-Rays and Radioactivity in Medicine:
With Emphasis on Photographic Records of the Early
Years, Richard F. Mould, ISBN 0-7503-0224-0, 1993,
236 pp, IOPP, Bristol (UK).
Essentials of Radiology Physics, Charles A. Kelsey,
ISBN: 0875273548, 1985, 467 pp, W.H. Green.
Introduction to Radiological Physics and Radiation
Dosimetry, Frank Herbert Attix, ISBN: 0-47101146-0, 1986, 640 pp, Advanced Medical Publishing,
Madison (WI).
Meandering in Medical Physics: A Personal Account
of Hospital Physics, J.E. Roberts, N.G. Trott,
ISBN: 0750304944, 1999, 181 pp, IOPP, Bristol
(UK).
Medical Physics and Biomedical Engineering, Brown
BH, Smallwood RH, Barber DC, Lawford PV; Hose
DR, ISBN: 0750303670, 1998, 768 pp, IOPP, Bristol
(UK).
Medical Physics Handbook of Units and Measures,
Freim J, Jr, ISBN: 0944838308, 1992, 47 pp, Medical
Physics Publishing, Madison (WI).
Medical Radiation Physics: Roentgenology, Nuclear Medicine & Ultrasound, Hendee WR, ISBN: 0815142404,
1979, 517 pp, Year Book Medical Publishers.
Physics in Medicine and Biology Encyclopedia (2
Volume Set), T.F. McAinsh, (editor), ISBN:
0080264972, 1986, Pergamon, Elmsfood (NY).
Physics and Engineering in Medicine in the New Millennium, Sharp PF, Perkins AC editors., ISBN:
0904181952, 2000, 156 pp, IPEM, York, (UK).
Physics of Radiology, 4th ed., Johns H E, John Robert
Cunningham, ISBN: 0398046697, 1983, 796 pp,
Charles C. Thomas.
Physics of Radiology, second edition, Wolbarst A B,
ISBN: 1-930524-22-6. Published: 2005, 660 pp, Medical Physics Publishing, Madison (WI).
Physics of the Body, Cameron JR, Skofronick JG, Grant
RM, ISBN: 094483891X, 1999, 394 pp, Medical Physics Publishing, Madison (WI).
Principles and Practice of Clinical Physics & Dosimetry,
Michael L.F. Lim, ISBN: 1-883526-11-6, 2005, 500
pp, Advanced Medical Publishing, Madison (WI).
Principles of Radiological Physics, 4th ed., Graham D,
Cloke P, ISBN: 0443070733, 2003, 576 pp, Churchill
Livingstone.
Radiation Biophysics, Alpen EL, ISBN: 0120530856,
1998, 484 pp, Academic Press.
Radiation Physics Handbook for Medical Physicists,
Ervin B. Podgorsak, ISBN: 3540250417, 2005, 360
pp, Springer, New York.
338
Review of Radiological Physics, Walter Huda, Richard

M. Slone, ISBN: 0781736757, 2002, 350 pp, Lippincott Williams & Wilkins, Philadelphia.
Topical
How the Body Works, Lenihan J, ISBN: 0944838-48-0,
1995, 200 pp, Medical Physics Publishing, Madison
(WI).
Physics of the Body 2nd ed., Cameron J, et al., ISBN: 0944838-91-X, 1999, 394 pp, Medical Physics Publishing, Madison (WI).
Medical Applications of Nuclear Physics, Bethge K,
Kraft G, Kreisler P, Walter G, ISBN: 3540208054,
2004, 208 pp, Springer.
Progress in Medical Radiation Physics, Orton CG,
ISBN: 0306417898, 1985, 248 pp, Plenum, New York.
Radiotherapy Physics: In Practice, Williams JR,

Thwaites DI, editors, ISBN: 0-19-262878-X, 2000,
362 pp, Oxford University Press, New York.
Review of Radiation Oncology Physics, Prasad SC,
ISBN: 1-930524-08-0, 2002, 95 pp, Medical Physics
Publishing, Madison (WI).
Study Guide for Radiation Oncology Physics Board
Exams, Berman B, ISBN: 0-944838-94-4, 2000, 112
pp, Medical Physics Publishing, Madison (WI).
The Physics of Radiation Therapy, Hardbound 3rd ed.,
Khan F, ISBN: 0-7817-3065-1, 2003, 511 pp, Wiley,
New York.
Walter & Millers Textbook of Radiotherapy Radiation
Physics, Therapy and Oncology, 6th ed., Bomford CK
et al., ISBN: 0443062013, 2003, 660 pp, Elsevier.
Topical
RADIATION ONCOLOGY PHYSICS
General
AAPM Monograph No. 15, Radiation Oncology Physics,
Kereiakes J, Elson H, Born C, editors, ISBN:
0-883185-33-4, 1986, 812 pp, Medical Physics Publishing, Madison (WI).
AAPM Monograph No. 26, General Practice of Radiation Oncology Physics in the 21st Century, Almon
Shiu & David Mellenberg, ISBN: 0-944838-98-7,
(WI).
Applied Physics for Radiation Oncology, Robert Stanton
& Donna Stinson ISBN: 0-944838-60-X, 1996, 375 pp,
Medical Physics Publishing, Madison (WI).
Biomedical Particle Accelerators, Scharf WH, Siebers
JV, ISBN: 1563960893, 1994, 480 pp, Springer.
Blackburns Introduction to Clinical Radiation Therapy
Physics, Benjamin Blackburn ISBN: 0-944838-06-5,
(WI).
Clinical Radiotherapy Physics, 2nd ed., Jayaraman S,
Lanzl LH, Lanzl EF, ISBN: 3540402845, 2004, 523
pp, Springer.
Handbook of Radiotherapy Physics: Theory and
Practice, Mayles P, Nahum A, Rosenwald J-C,
ISBN: 0750308605, 2005, 700 pp, IOPP, Bristol
(UK).
Modern Technology of Radiation Oncology, Van Dyk J,
Practical Radiotherapy: Physics and Equipment,
Cherry P, Duxbury A, ISBN: 1900151065, 1998,
224 pp, Cambridge University Press, New York.
Radiation Therapy Physics, Hendee WR, Ibbott GS,
Hendee EG, ISBN: 0471394939, 2004, 450 pp, Wiley,
New York.
Radiotherapy Physics and Equipment, Morris S,
Williams A, ISBN: 0443062110, 2001, 176 pp,
Churchill Livingstone.
3-D Conformal and Intensity Modulated Radiation

Therapy: Physics and Clinical Applications, Purdy
JA, Grant W III, Palta JR, Butler EB, Perez CA,
editors, ISBN: 1-883526-10-8, 2001, 650 pp,
Advanced Medical Publishing, Madison (WI).
A Practical Guide to 3-D Planning and Conformal
Radiation Therapy, Purdy JA, Starkschall G, ISBN:
1-883526-07, 1999, 400 pp, Advanced Medical Publishing, Madison (WI).
A Practical Guide to Intensity-Modulated Radiation
Therapy, Memorial Sloan-Kettering Cancer Center,
ISBN: 1-930524-13-7, 2003, 450 pp. Medical Physics
AAPM Manual No. 2: Workbook on Dosimetry and
Treatment Planning for Radiation Oncology Residents, Wu RK, Gerbi BJ, Doppke KP, editors, ISBN:
0-88318-916-X, 1991, 32 pp, Medical Physics Publishing, Madison (WI).
AAPM Monograph No. 2, Practical Aspects of Electron
Beam Treatment Planning, Orton CG, Bagne F,
editors, ISBN: 0-88318-247-5, 1978, 109 pp, Medical
AAPM Monograph No. 7, Recent Advances in
Brachytherapy Physics, Shearer DR, editors, ISBN:
AAPM Monograph No. 8, Physical Aspects of
Hyperthermia, Nussbaum GH, editors, ISBN: 088318-414-1, 1982, 656 pp, Medical Physics Publishing, Madison (WI).
AAPM Monograph No. 9, Advances in Radiation Therapy Treatment Planning, Wright AE, Boyer A, editors, ISBN: 0-883184-23-0, 1982, 626 pp, Medical
AAPM Monograph No. 16, Biological, Physical and Clinical Aspects of Hyperthermia, Paliwal BR, Hetzel FW,
Dewhirst M, editors, ISBN: 0-88318-558-X, 1988, 483
AAPM Monograph No. 28, Intravascular Brachytherapy/Fluoroscopically Guided Interventions, Balter S,
Chan RC, Shope TB Jr, editors, ISBN: 1-930524-10-2,

(WI).
AAPM Monograph No. 29, Intensity-Modulated Radiation Therapy: The State of the Art, Palta JR, Rockwell Mackie T, editors, ISBN: 1-930524-16-1, 2003,
904 pp. Medical Physics Publishing, Madison (WI).
AAPM Proceedings No. 2, Proceedings of the Symposium on Electron Dosimetry and Arc Therapy,
Paliwal BR, editor, ISBN: 0-88318-404-4, 1981, 384
AAPM Proceedings No. 3, Proceedings of a Symposium
on Quality Assurance of Radiotherapy Equipment,
Starkschall G, editor, ISBN: 0-88318-422-2, 1982,
242 pp, Medical Physics Publishing, Madison (WI).
AAPM Proceedings No. 5, Optimization of Cancer
Radiotherapy, Paliwal BR, Herbert DE, Orton CG,
AAPM Proceedings No. 12, Biological & Physical Basis
of IMRT & Tomotherapy, Paliwal BR, Herbert DE,
Fowler JF, Mehta M, editors, ISBN: 1-930524-11-0,
(WI).
AAPM Report No. 7, Protocol for Neutron Beam Dosimetry, Radiation Therapy Committee Task Group
18; ISBN: 0-88318-276-9, 1980, 51 pp, Medical Physics Publishing, Madison (WI).
AAPM Report No. 13, Physical Aspects of Quality Assurance in Radiation Therapy, Radiation Therapy Committee Task Group 24, with contribution from Task
Group 22, ISBN: 0-88318-457-5, 1984, 63 pp, Medical
AAPM Report No. 17, The Physical Aspects of Total and
a Half Body Photon Irradiation, Radiation Therapy
Committee Task Group 29; ISBN: 0-88318-513-X,
(WI).
AAPM Report No. 21, Specification of Brachytherapy
Source Strength, Radiation Therapy Committee
Task Group 32; ISBN: 0-88318-545-8, 1987, 21 pp,
AAPM Report No. 23, Total Skin Electron Therapy:
Technique and Dosimetry Radiation Therapy Committee Task Group 30; ISBN: 0-88318-556-3, 1987, 55
AAPM Report No. 24, Radiotherapy Portal Imaging
Quality, Radiation Therapy Committee Task Group
AAPM Report No. 26, Performance Evaluation of
Hyperthermia Equipment, Hyperthermia Committee Task Group 1; ISBN: 0-88318-636-5, 1989, 46
AAPM Report No. 27, Hyperthermia Treatment Planning, Hyperthermia Committee Task Group 2; ISBN:
339
AAPM Report No. 32, Clinical Electron-Beam Dosimetry, Radiation Therapy Committee Task Group 25,
AAPM Report No. 40, Radiolabeled Antibody Tumor
Dosimetry (Reprinted from Medical Physics, Vol.
20, Issue 2), Nuclear Medicine Committee Task
Group 2; ISBN: 1-56396-233-0, 1993, 112 pp, Medical
AAPM Report No. 41, Remote Afterloading Technology,
Remote Afterloading Technology Task Group 41;
AAPM Report No. 45, Management of Radiation
Oncology Patients with Implanted Cardiac Pacemakers (Reprinted from Medical Physics, Vol. 21,
Issue 1), Task Group 34. ISBN: 1-56396-380-9,
(WI).
AAPM Report No. 46, Comprehensive QA for Radiation
Oncology (Reprinted from Medical Physics, Vol. 21,
Issue 4), Radiation Therapy Committee Task Group
AAPM Report No. 47, AAPM Code of Practice for Radiotherapy Accelerators (Reprinted from Medical Physics, Vol. 21, Issue 4), Radiation Therapy Task Group
AAPM Report No. 48, The Calibration and Use of PlaneParallel Ionization Chambers for Dosimetry of Electron Beams (Reprinted from Medical Physics, Vol. 21,
Issue 8) Radiation Therapy Committee Task Group
AAPM Report No. 50, Fetal Dose from Radiotherapy
with Photon Beams (Reprinted from Medical Physics, Vol. 22, Issue 1), Radiation Therapy Committee
Task Group 36; ISBN: 1-56396-453-8, 1995, 20 pp,
AAPM Report No. 54, Stereotactic Radiosurgery, Radiation Therapy Committee Task Group 42; ISBN:
AAPM Report No. 55, Radiation Treatment Planning
Dosimetry Verification, AAPM ISBN: 1-56396-534-8,
(WI).
AAPM Report No. 56, Medical Accelerator Safety Considerations (Reprinted from Medical Physics, Vol. 20,
AAPM Report No. 59, Code of Practice for Brachytherapy Physics (Reprinted from Medical Physics,
Vol. 24, Issue 10), Radiation Therapy Committee Task
Group 56; ISBN: 1-888340-14-2, 1997, 42 pp, Medical
340
AAPM Report No. 61, High Dose-Rate Brachytherapy

Treatment Delivery (Reprinted from Medical Physics, Vol. 25, Issue 4), Radiation Therapy Committee
Task Group 59; ISBN: 1-888340-17-7, 1998, 29
AAPM Report No. 62, Quality Assurance for Clinical
Radiotherapy Treatment Planning (Reprinted
from Medical Physics, Vol. 25, Issue 10), Radiation
Therapy Committee Task Group 53; ISBN:
1-888-340-18-5, 1998, 57 pp, Medical Physics Publishing, Madison (WI).
AAPM Report No. 66, Intravascular Brachytherapy
Physics (Reprinted from Medical Physics, Vol. 26,
AAPM Report No. 67, Protocol for Clinical Reference
Dosimetry of High-Energy Photon and Electron
Beams (Reprinted from Medical Physics, Vol. 26,
Issue 9) Task Group 51; ISBN: 1-888340-25-8,
(WI).
AAPM Report No. 68, Permanent Prostate Seed
Implant Brachytherapy (Reprinted from Medical
Physics, Vol. 26, Issue 10), Task Group 64; ISBN:
AAPM Report No. 69, Recommendations of the AAPM
on 103Pd Interstitial Source Calibration and Dosimetry: Implications for Dose Specification and Prescription, AAPM, ISBN: 1-888340-27-4, 2000, 9 pp,
AAPM Report No. 71, A Primer for Radioimmunotherapy and Radionuclide Therapy Task Group 7, ISBN:
AAPM Report No. 72, Basic Applications of Multileaf
Collimators, Task Group 50 ISBN: 1-888340-30-4,
(WI).
AAPM Report No. 75, Clinical Use of Electronic Portal
Imaging, Radiation Therapy Committee Task Group
58, ISBN: 1-888340-34-7, 2001, 26 pp, Medical Physics Publishing, Madison (WI).
AAPM Report No. 76, AAPM Protocol for 40-300 kV
X-ray Beam Dosimetry in Radiotherapy and Radiobiology, AAPM, ISBN: 1-888340-35-5, 2001, 26 pp,
AAPM Report No. 81, Dosimetric Considerations for
Patients with Hip Prostheses Undergoing Pelvic
Irradiation, Radiation Therapy Committee Task
Group 63 (Reprinted from Medical Physics, Vol. 30,
Issue 6), ISBN: 1-888340-42-8, 2003, 21 pp, Medical
AAPM Report No. 82, Guidance Document on Delivery,
Treatment Planning, and Clinical Implementation of
IMRT, AAPM Radiation Therapy Committee, ISBN:
AAPM Report No. 83, Quality Assurance for ComputedTomography Simulators and the ComputedTomography-Simulation Process, Radiation Therapy
Committee Task Group 66, ISBN: 1-888340-44-4,
(WI).
AAPM Report No. 84, A Revised AAPM Protocol for
Brachytherapy Dose Calculations (Reprinted from
Medical Physics, Vol. 31, Issue 3, pp. 633-674), Radiation Therapy Committee, ISBN: 1-888340-46-0, 2004,
AAPM Report No. 85, Tissue Inhomogeneity Corrections for Megavoltage Photon Beams, Task Group
No. 65 of the Radiation Therapy Committee, ISBN: 1888340-47-9, 2004, 124 pp, Medical Physics Publishing, Madison (WI).
AAPM Report No. 86, Quality Assurance for Clinical
Trials: A Primer for Physicists, Subcommittee on
Quality Assurance Physics for Cooperative Trials
of the Radiation Therapy Committee, ISBN: 188340-48-7, 2004, 68 pp, Medical Physics Publishing,
Madison (WI).
Achieving Quality in Brachytherapy, BR Thomadsen,
ISBN: 0750305541, 1999, pp, 268, Advanced Medical
A Practical Guide to CT-Simulation, Edited by: Coia L,
Schultheiss T, Hanks G, ISBN: 1-883526-04-3, 1995,
216 pp, Advanced Medical Publishing, Madison (WI).
Brachytherapy Physics (1994 AAPM Summer School),
Williamson J et al., ISBN: 0-944838-50-2, 1995, 715
Clinical Target Volumes in Conformal and Intensity Modulated Radiation Therapy, Gregorie V,
Scalliet and P, Ang KK, ISBN: 3540413804, 2003,
300 pp, Springer Verlag.
Contemporary IMRT: Developing Physics and Clinical
Implementation, Webb S, ISBN: 0750310049, 2004,
CT Simulation for Radiotherapy, Jani S, ISBN:
Geometric Uncertainties in Radiotherapy, BIR, 2003,
ISBN: 0905749537, The British Institute of Radiology.
Intraoperative Irradiation, Techniques and Results,
Gunderson LL, Wilett CG, Harrison LB, Calvo FA,
editors ISBN: 0-89603-523-9, 1999, 560 pp, Advanced
Medical Publishing, Madison (WI).
Intraoperative Radiation Therapy, Ralph Dobelbower,
Jr. and Mitsuyuki Abe, ISBN: 0849368464, 1989, 432
pp, CRC Press, Boca Raton (FL).
Introduction to Clinical Radiation Oncology, 3rd ed.,
Coia L, Moylan D, ISBN: 0-944838-70-7, Published:
March 1998, 568 pp, Medical Physics Publishing,
Madison (WI).
IPEM Report No. 68, A Guide to Commissioning &
Quality Control of Treatment Planning Systems,
Shaw J, editor, ISBN: 0904181839, 1996, IPEM, York
(UK).
IPEM Report No. 81, Physics Aspects of Quality Control

in Radiotherapy, Mayles WPM, Lake RA, McKenzie
AL, Macaulay EM, Morgan HM, Powley SK, ISBN:
0904181928, 1998, IPEM, York (UK).
IPEM Report No. 83, Targeted Radiotherapy, Fleming
JS, Perkins AC, ISBN: 0904181979, Published: 2000,
112 pp, IPEM, York (UK).
Linac and Gamma Knife Radiosurgery, Isabelle M.
Germano, ISBN: 1879284707, 2000, 295 pp,
Linear Accelerators for Radiation Therapy, D. Greene,
ISBN: 0750304766, 1997, 288 pp, IOPP, Bristol (UK).
Monitor Unit Calculations for External Photon and
Electron Beams, Gibbon JP, editor, ISBN: 1883526-08-6, 2000, 152 pp, Advanced Medical Publishing, Madison (WI).
Physical Aspects of Brachytherapy, Godden TJ, ISBN:
0852745117, 1988, 304 pp, IOPP, Bristol (UK).
Physical Aspects of Stereotactic Radiosurgery, Phillips
M H, ISBN: 0306445352, 1993, 286 pp, Plenum, New
York.
Physics and Technology of Hyperthermia, Field SB,
Franconi C, ISBN: 9024735092, 1999, 668 pp,
Springer.
Physics of Electron Beam Therapy, Klevenhagen SC,
Physics of Radiotherapy X-Rays from Linear Accelerators, Metcalfe P et al., ISBN: 0-944838-76-6, 1997,
Practical Essentials of Intensity Modulated Radiation
Therapy, Chao KSC, Smith Apisarnthanarax, and
Gokhan Ozyigit, ISBN: 0-7817-5279-5, 2004, 324 pp,
Practical Manual of Brachytherapy, Pierquin B,
Marinello G, ISBN: 0-944838-73-1, 1997, 296 pp,
Primer on Theory and Operation of Linear Accelerators,
2nd ed., Karzmark CJ, Morton R, ISBN: 0-94483866-9, 1998, 50 pp, Medical Physics Publishing,
Madison (WI).
Principles and Practice of Brachytherapy, Nag S, editor,
ISBN: 0879936541, 1997, 752 pp, Futura.
Principles and Practice of Brachytherapy Using Afterloading Systems, Joslin CA, Flynn A, Hall EJ, editors, ISBN: 0-340-74209-7, 2001, 464 pp, Edward
Arnold, London.
Protocol and Procedures for Quality Assurance of Linear
Accelerators, Constantinou, ISBN: 0-9638266-0-3,
(WI).
Quality
Assurance
in
Radiotherapy
Physics,
Starkschall G, ISBN: 0944838219, 1991, 387 pp,
Radiation Therapy Planning, Bentel GC, ISBN:
0070051151, 1995, 643 pp, McGraw-Hill, New York.
Radiotherapy In Practice Brachytherapy, Hoskin PJ,
Coyle C, ISBN: 0198529406, 2005, 224 pp, Oxford
University Press, New York.
341
Study Guide for Radiation Oncology Physics Board

Exams, Berman B, Thomadsen B, ISBN: 0-94483894-4, 2000, 112 pp, Medical Physics Publishing,
Madison (WI).
The Physics and Radiobiology of Fast Neutron Beams,
Bewley DK, ISBN: 085274093x, 1989, 192 pp, IOPP,
Bristol (UK).
The Physics of Conformal Radiotherapy: Advances in
Technology, Webb S, ISBN: 0750303972, 1997, 382
pp, IOPP, Bristol (UK).
The Physics of Modern Brachytherapy for Oncology,
Baltas D, Kreiger H, Zamboglou N, ISBN:
The Physics of Three Dimensional Radiation Therapy:
Conformal Radiotherapy, Radiosurgery and Treatment Planning, Webb S, ISBN: 075030247x, 1993,
The Q Book The Physics of Radiotherapy X-Rays:
Problems and Solutions Metcalfe P et al., ISBN: 0944838-86-3, 1998, 100 pp, Medical Physics Publishing, Madison (WI).
The Theory & Practice of Intensity Modulated Radiation
Therapy, Sternick S, editor, ISBN: 1-883526-05-1,
1997, 256 pp, Advanced Medical Publishing, Madison
(WI).
The Use of Computers in Radiation Therapy: Schlegel
W, Bortfeld T, editors, ISBN: 3540671765, 2000, 604
pp, Springer, New York.
The Use of Plane Parallel Ionization Chambers in High
Energy Electron and Photon Beams: An International Code of Practice for Dosimetry, IAEA, ISBN:
9201048963, 1997, 125 pp, IAEA.
Therapy Physics Review, Paliwal B, ISBN: 0-944838-677, 1996, 65 pp, Medical Physics Publishing, Madison
(WI).
Three-Dimensional Radiation Treatment: Technological Innovations and Clinical Results, Kneschaurek
P, Molls M, Feldmann HJ, ISBN: 3805569475, 2000,
S. Karger.
Topics in Dosimetry & Treatment Planning for Neutron
Capture Therapy, Zamenhof RG, Solares GR,
Harling OK, editors, ISBN: 1-883526-02-7, 1994,
245 pp, Advanced Medical Publishing, Madison (WI).
Treatment Planning in Radiation Oncology, Khan F M,
Potish R, editors, ISBN: 0-683-04607-1, 1997, 608 pp,
Lippincott, New York.
DIAGNOSTIC RADIOLOGICAL PHYSICS

General
AAPM Monograph No. 3, The Physics of Medical Imaging: Recording System Measurements and Techniques (1979 Summer School), Haus AG, editor, ISBN:
0-88318-260-2, 624 pp, Medical Physics Publishing,
Madison (WI).
AAPM Monograph No. 23, The Expanding Role of Medical Physics in Diagnostic Imaging, Frey GD,
342
Sprawls P, editors, ISBN: 1-888340-09-6, 1997, 583

Christensens Physics of Diagnostic Radiology, Curry T
S III, Dowdey JE, Murry RC Jr, ISBN: 0812113101,
1990, 522 pp, Lippincott, New York.
IPEM Report 61, Physics in Diagnostic Radiology,
Faulkner K, Cranley K, Starritt HC, Wankling
PF, editors, ISBN: 090418160X, 1990, 150 pp,
IPEM, York (UK).
Physics for Diagnostic Radiology, Dendy P P, Heaton B,
Practical Radiography, Robert Ward, ISBN: 0-94483849-9, (1996 reprint), 112 pp, Medical Physics Publishing, Madison (WI).
The Physics of Diagnostic Imaging, Dowsett DJ,
Johnston RE, Kenny PA, ISBN: 0412460602, 1998,
609 pp, Edward Arnold, London.
Topical
AAPM Monograph No. 4, Quality Assurance in Diagnostic Radiology, Waggener R, Wilson C, editors,
AAPM Monograph No. 30, Specifications, Performance
Evaluation and Quality Assurance of Radiographic
and Fluoroscopic Systems in the Digital Era,
Goldman L, Yester M, editors, ISBN: 1-930524-218, 2004, 300 pp, Medical Physics Publishing, Madison
(WI).
AAPM Report No. 4, Basic Quality Control in Diagnostic
Radiology, Task Force On Quality Assurance
Protocol; ISBN: 0-88318-251-3, 1977, 57 pp, Medical
AAPM Report No. 14, Performance Specifications and
Acceptance Testing for X-Ray Generators and
Automatic Exposure Control Devices, AAPM, ISBN:
AAPM Report No. 15, Performance Evaluation and
Quality Assurance in Digital Subtraction Angiography, Diagnostic X-Ray Imaging Committee/
DigitalRadiography/Fluorography Task Group;
AAPM Report No. 29, Equipment Requirements and
Quality Control for Mammography, Diagnostic
X-Ray Imaging Committee Task Group 7, ISBN: 088318-807-4, 1990, 72 pp, Medical Physics Publishing, Madison (WI).
AAPM Report No. 31, Standardized Methods for Measuring Diagnostic X-Ray Exposures, Diagnostic
X-Ray Imaging Committee Task Group 8, ISBN: 088318-874-0, 1990, 22 pp, Medical Physics Publishing, Madison (WI).
AAPM Report No. 58, Managing the Use of Fluoroscopy
in Medical Institutions, Radiation Protection
Committee Task Group 6; ISBN: 1-888340-13-4,

(WI).
AAPM Report No. 60, Instrumentation Requirements of
Diagnostic Radiological Physicists, Diagnostic X-Ray
Imaging Committee Task Group 4; ISBN: 1-88834015-0, 1998, 40 pp, Medical Physics Publishing,
Madison (WI).
AAPM Report No. 70, Cardiac Catheterization Equipment Performance, Diagnostic X-ray Imaging
Committee, Task Group 17, ISBN: 1-888340-28-2,
(WI).
AAPM Report No. 74, Quality Control in Diagnostic
Radiology, Task Group 12 ISBN: 1-888340-33-9,
(WI).
Advances in Film Processing Systems Technology and
Quality Control in Medical Imaging, Haus AG, ISBN:
1-930524-01-3, 2001, 245 pp., Medical Physics Publishing, Madison (WI).
Basics of Film Processing in Medical Imaging, Haus A,
Jaskulski S, ISBN: 0-944838-78-2, 1997, 338 pp,
Digital Mammography Proceedings, Yaffe M, ISBN:
1-930524-00-5, 2001, 856 pp Medical Physics Publishing, Madison (WI).
Interventional Fluoroscopy: Physics, Technology,
Safety, Balter S, ISBN: 0471390100, 2001, 284 pp,
Wiley, New York.
IPEM Report No. 32, Measurement of the Performance
Characteristics of Diagnostic X-Ray Systems Used in
Medicine. Part 1: X-Ray Tubes and Generators,
Cranley K ISBN: 090418174X, 1995, 28 pp, IPEM,
York (UK).
Medicine. Part II: X-Ray Image Intensifier Television
Systems, Starritt H C, ISBN: 0904181758, 1996, 61
pp, IPEM, York (UK).
Medicine. Part IV: X-Ray Intensifying Screens,
Films, Processors and Automatic Exposure Control
Systems, Holubinka M R, ISBN: 0904181774, 1996,
Medicine. Part V: Conventional Tomographic Equipment, ISBN: 0904181782, 1996, 18 pp, IPEM, York
(UK).
Medicine. Part VI: X-Ray Image Intensifier Fluorography Systems, Robertson J, ISBN: 0904181790,
1995, 21 pp, IPEM, York (UK).
IPEM Report No. 59, The Commissioning & Routine
Testing of Mammographic X-Ray Systems 2nd ed.,
Law J, Dance DR, Faulkner K, Fitzgerald MC,
Ramsdale ML, Robinson A, ISBN: 0904181723, 1994,

IPEM Report No. 67 Quality Assurance in Dental
Radiology, Starritt HC, Faulkner K, Wankling PF,
Cranley K, Robertson J, Young K, ISBN:
0904181677, 1994, 25 pp, IPEM, York (UK).
IPEM Report No. 77, Recommended Standards for
Routine Testing of Diagnostic X-Ray Imaging Systems, ISBN: 0904181871, 1997, 64 pp, IPEM, York
(UK).
IPEM Report No. 78, Catalogue of Diagnostic X-Ray
Spectra & Other Data Cranley K, Gilmore BJ,
Fogarty GWA, Desponds L, ISBN: 090418188X,
1997, IPEM, York (UK).
IPEM Report No. 79, The Critical Examination of X-Ray
Generating Equipment in Diagnostic Radiology,
ISBN: 0904181898, 1998, 17 pp, IPEM, York (UK).
Mammography Quality Control: The Why & How Book,
Calvin Myers, ISBN: 0-944838-83-9, 1997, 43 pp,
Medical Imaging and Radiation Protection for Medical
Students and Clinical Staff, Martin CJ, Dendy PP,
Corbett RH, 2003, The British Institute of Radiology.
Practical Digital Imaging and PACS (1999 AAPM Summer School), Seibert A et al., ISBN: 0-944838-92-8,
(WI).
Principles of Radiographic Imaging: An Art and a
Science, Carlton RR, Adler A, ISBN: 0766813002,
2000, 752 pp, Thomson Delmar Learning.
Radiation Exposure and Image Quality in X-Ray Diagnostic Radiology: Physical Principles and Clinical
Applications, Aichinger H, Dierker J, Joite-Barfu
S, Sa bel M, ISBN: 3540442871, 2003, 212 pp,
Springer, New York.
Radiology Review: Radiologic Physics, Nickoloff EL,
Naveed Ahmad, ISBN: 1416022600, 2005, 272 pp,
Saunders, Philadelphia.
Screen Film Mammography: Imaging Considerations
and Medical Physics Responsibilities, Gary Barnes
and G. Donald Frey, ISBN: 0-944838-12-X, 1991, 127
IMAGING
General
3D Imaging in Medicine, 2nd ed., Udupa JK, Herman
GT, ISBN: 084933179X, 1999, 384 pp, CRC Press,
Boca Raton (FL).
Essentials of Diagnostic Imaging, Guebert G M, Pirtle
OL, Yochum TR, ISBN: 0801674557, 1995, 252 pp,
Mosby, Philadelphia.
Foundations of Image Science, Barrett HH, Myers K,
ISBN: 0471153001, 2003, 1100 pp, Wiley, New York.
Fundamentals of Medical Imaging, Paul Suetens, ISBN:
0521803624, 2002, 294 pp, Cambridge University
Press, New York.
343
Handbook of Medical Imaging, Volume 1: Physics and

Psychophysics, Beutel J, Kundel HL, Van Metter RL,
ISBN: 0819436216, 2000, 968 pp, SPIE.
Introduction to Biomedical Imaging, Webb AG, ISBN:
0471237663, 2002, 264 pp, Wiley, New York.
Introduction To The Principles of Medical Imaging, Guy
C, ISBN: 1860945023, 2005, 400 pp, Imperial College
Press, London.
Medical Imaging 2004: Physics of Medical Imaging
(SPIE Proceedings), Yaffe MJ, ISBN: 0819452815,
2004, SPIE.
Medical Imaging Physics, 4th ed., Hendee WR, Ritenour
ER, ISBN: 0-471-38226-4, 2002, 536 pp, Wiley, New
York.
Physics for Medical Imaging, Farr RF, Allisy-Roberts
PJ, ISBN: 0702017701, 1997, 276 pp, Bailliere
Tindall.
Physics of Medical Imaging, Dobbins JT, Boone JM,
ISBN: 0819427810, 1998, 842 pp, SPIE.
Principles of Imaging Science and Protection, Thompson MA, Hattaway MP, Hall JD, ISBN: 0721634281,
1994, 522 pp, Saunders, Philadelphia.
Principles of Medical Imaging, Kirk Shung K, Smith
MB, Tsui BMW, ISBN: 0126409706, 1992, 289 pp,
Academic Press.
The Essential Physics of Medical Imaging, Hardbound,
2nd ed., Bushberg JT, Seibert JA, Leidholdt EM Jr,
Boone JM, ISBN: 0-683-30118-7, 2001, 965 pp,
Lippincott.
The Physics of Diagnostic Imaging, 2nd ed., Dowsett DJ,
Kenny PA, Johnston RE, ISBN: 0412460602, 2005,
Hodder Headline (Arnold).
The Physics of Medical Imaging, Webb S, ISBN:
Topical
AAPM Monograph No. 11, Electronic Imaging in Medicine, Fullerton GD, Hendee W, Lasher J, Properzio
W, Riederer S, editors, ISBN: 0-88318-454-0, 1983,
AAPM Monograph No. 12, Recent Developments in
Digital Imaging (1984 Summer School), Doi K, Lanzl
L, Lin P-J P, editors, ISBN: 0-88318-463-X, 1984,
AAPM Monograph No. 25, Practical Digital Imaging
and PACS (1999 AAPM Summer School), Seibert A
et al., ISBN: 0-944838-92-8, 1999, 577 pp, Medical
Electrical Impedance Tomography: Methods, History
and Applications, Holder DS, ISBN: 0750309520,
2005, 456 pp, IOPP, Bristol (UK).
Mathematics of Medical Imaging, Epstein CL, ISBN:
0130675482, 2003, 768 pp, Prentice Hall, New York.
Medical Imaging Signals and Systems, Prince JL, Links
J, ISBN: 0130653535, 2005, 550 pp, Prentice Hall,
New York.
344
Wavelet Analysis with Applications to Image Processing, Lakshman Prasad and S. Sitharama Iyengar,
ISBN: 0849331692, 1997, 304 pp, CRC Press, Boca
Raton (FL).
COMPUTERIZED TOMOGRAPHY
General
AAPM Monograph No. 6, Medical Physics of CT and
Ultrasound: Tissue Imaging and Characterization
(1980 Summer School), Fullerton GD, Zagzebski J,
Computed Tomography: Fundamentals, System Technology, Image Quality, Applications, Kalender WA,
ISBN: 3-8957-8081-2, 2000, 220 pp, Advanced Medical Publishing, Madison (WI).
CT Physics: The Basics, Villafana T, ISBN: 0683307118,
2002, 250 pp, Lippincott.
Topical
AAPM Report No. 1, Phantoms for Performance Evaluation and Quality Assurance of CT Scanners,
AAPM, ISBN: 1-888340-04-5, 1977, 23 pp, Medical
AAPM Report No. 39, Specification and Acceptance
Testing of Computed Tomography Scanners, Diagnostic X-Ray Imaging Committee Task Group 2;
Medicine. Part III: Computed Tomography X-Ray
Scanners, ISBN: 0904181766, 2003, 94 pp, IPEM,
York (UK).
NUCLEAR MEDICINE
General
AAPM Monograph No. 10, Physics of Nuclear Medicine:
Recent Advances (1983 Summer School), Rao D,
Chandra R, Graham M, editors, ISBN: 0-88318440-0, 1983, 560 pp, Medical Physics Publishing,
Madison (WI).
Diagnostic Nuclear Medicine: A Physics Perspective,
Hamilton DI, ISBN: 3540006907, 2004, 465 pp,
Springer, New York.
Essentials of Nuclear Medicine Physics, Powsner RA,
Powsner ER, ISBN: 0632043148, 1998, 199 pp,
Blackwell, Cambridge (MA).
Handbook of Nuclear Medicine, Madsen M, Ponto J,
Introductory Physics of Nuclear Medicine, Ramesh
Chandra, ISBN: 0812114426, 1992, 221 pp, Lea &
Febiger, Philadelphia (PA).
Nuclear Medicine and PET: Technology and Techniques, Christian PE, Bernier D, Langan JK, ISBN:
0323019641, 2003, 640 pp, Mosby.
Nuclear Medicine Physics: The Basics, Ramesh Chandra,
ISBN: 068330092X, 1998, 182 pp, Lippincott, New
York.
Physics in Nuclear Medicine, Cherry SR, Sorenson J,
Phelps M, ISBN: 072168341X, 2003, 523 pp, Saunders, Philadelphia.
Practical Nuclear Medicine, 3rd ed., Sharp PF, Gemmell
HG, Murray AD, ISBN: 185233875X, 2005, 352 pp,
Springer, New York.
Principles and Practice of Nuclear Medicine, Early PJ,
Bruce D,. Sodee MD, ISBN: 0801625777, 1995,
877 pp, Mosby.
Topical
AAPM Manual No. 1: Nuclear Medicine Instrumentation Laboratory Exercises for Radiology Residency
Training, Van Tuinen R J, Grossman LW, Kereiakes
JG, editors, ISBN: 0-88318-0001, 1994, 81 pp, Medical Physics Publishing, Madison (WI).
AAPM Monograph No. 1, Biophysical Aspects of Medical
Use of Technetium-99m, Kereiakes JG, Corey KR,
AAPM Monograph No. 18, Expanding the Role of Medical Physics in Nuclear Medicine (1989 Summer
School), Frey GD, Yester MV, editors, ISBN:
AAPM Report No. 6, Scintillation Camera Acceptance
Testing & Performance Evaluation, Nuclear Medicine Committee; ISBN: 0-88318-275-0, 1980, 23 pp,
AAPM Report No. 9, Computer-Aided Scintillation
Camera Acceptance Testing, Nuclear Medicine
Committee Task Group; ISBN: 0-88318-407-9,
(WI).
AAPM Report No. 22, Rotating Scintillation Camera
SPECT Acceptance Testing and Quality Control,
Nuclear Medicine Committee Task Group; ISBN:
AAPM Report No. 52, Quantitation of SPECT Performance (Reprinted from Medical Physics, Vol. 2, Issue
4), Nuclear Medicine Committee Task Group 4;
IPEM Report No. 65, Quality Standards in Nuclear
Medicine, Edited by Hart GC and Smith AH, ISBN:
0904181642, 1992, 123 pp, IPEM, York (UK).
IPEM Report No. 66, Quality Control of Gamma Cameras & Associated Computer Systems, Hannan J,
editor, ISBN: 0904181650, 1992, 62 pp, IPEM, York
(UK).
IPEM Report No. 85, Radioactive Sample Counting

Principles and Practice, Driver I, editor, ISBN:
0904181995, 2002, 63 pp, IPEM, York (UK).
IPEM Report No. 86, Quality Control of Gamma Camera
Systems, Bolster A, ISBN: 1903613132, 2003, 130 pp,
IPEM, York (UK).
IPEM Report No. 87, Basics of Gamma Camera Positron
Emission Tomography Hillel P, editor, ISBN:
1903613183, 2004, 73 pp, IPEM, York (UK).
Positron Emission Tomography: Basic Sciences,
Bailey DL, Townsend DW, Valk PE, Maisey MN,
ISBN: 1852337982, 2005, 382 pp, Springer, New
York.
Principles and Practice of Positron Emission Tomography, Wahl RL, ISBN: 0781729041, 2002, 442 pp,
Lippincott, New York.
Therapeutic Applications of Monte Carlo Calculations
in Nuclear Medicine, Habib Zaidi, ISBN:
MAGNETIC RESONANCE IMAGING AND SPECTROSCOPY
General
AAPM Monograph No. 14, NMR in Medicine: The
Instrumentation and Clinical Applications (1985
Summer School), Thomas SR, Dixon RL, editors,
ISBN: 0-88318-497-4, 1985 595 pp, Medical Physics
AAPM Monograph No. 21, The Physics of MRI, Michael
Bronskill, Sprawls P, editor, ISBN: 1-563962-05-5,
(WI).
Magnetic Resonance Imaging: Physical Principles and
Sequence Design, Haacke EM et al., ISBN:
0471351288, 1999, 914 pp, John Wiley & Sons,
Inc., New York.
Magnetic Resonance Imaging: Principles, Methods, and
Techniques, Sprawls P, ISBN: 0-944838-97-9, 2000,
Magnetic Resonance Imaging: Theory and Practice,
Vlaardingerbroek MT, Den Boer JA, ISBN:
3540600809, 1996, 347 pp, Springer, New York.
Magnetic Resonance in Medicine, 4th ed., Peter Rinck,
ISBN: 0632059869, 2001, 245 pp, Blackwell.
Non-Mathematical Approach to Basic MRI, Smith H,
Ranallo F, ISBN: 0-944838-02-2, Published: 1989,
Topical
AAPM Report No. 20, Site Planning for Magnetic Resonance Imaging Systems, Nuclear Magnetic Resonance Committee Task Group 2; ISBN: 0-88318530-X, 1986, 60 pp, Medical Physics Publishing,
Madison (WI).
AAPM Report No. 28, Quality Assurance Methods and
Phantoms for Magnetic Resonance Imaging (Reprinted from Medical Physics, Vol. 17, Issue 2),
345
Nuclear Magnetic Resonance Committee Task Group

1, ISBN: 0-88318-800-7, 1990, 9 pp, Medical Physics
AAPM Report No. 34, Acceptance Testing of Magnetic
Resonance Imaging Systems (Reprinted from Medical Physics, Vol. 19, Issue 1), Nuclear Magnetic
Resonance Committee Task Group 6; ISBN:
AAPM Report No. 77, Practical Aspects of Functional
MRI, Nuclear Medicine Committee Task Group 8,
AAPM Report No. 78, Proton Magnetic Resonance Spectroscopy in the Brain, Magnetic Resonance Task
Group 9, ISBN: 1-888340-38-X, 2002, 21 pp, Medical
Handbook of MRI Pulse Sequences, Matt Bernstein,
Kevin King, Xiaohong Joe Zhou, ISBN:
0120928612, 2004, 1040 pp, Academic Press.
Introduction to Functional Magnetic Resonance Imaging: Principles and Techniques, Buxton RB, ISBN:
Press, New York.
IPEM Report No. 80, Quality Control in Magnetic Resonance Imaging, RA Lerski, J De Wilde, D Boyce and J
Ridgeway, ISBN: 0904181901, 1999, 50 pp, IPEM,
York (UK).
Principles of Magnetic Resonance Imaging: A Signal
Processing Perspective, Liang Z-P, Lauterbur PC,
ULTRASOUND PHYSICS
General
AAPM Monograph No. 6, Medical Physics of CT and
Ultrasound: Tissue Imaging and Characterization
(1980 Summer School), Fullerton GD, Zagzebski J,
Advances in Ultrasound Techniques and Instrumentation, Wells PNT, ISBN: 0443088535, 1993, 192 pp,
Saunders, Philadelphia.
Clinical Ultrasound Physics: A Workbook for Physicists,
Residents, and Students, Kofler JMJr, et al., ISBN:
1-930524-06-4a, 2001, 85 pp, Medical Physics Publishing, Madison (WI).
Diagnostic Ultrasound: Physics and Equipment,
Hoskins P, Thrush A, Martin K, Whittingam T,
Hoskins PR, Thrush A, Whittingham T, ISBN:
Press, New York.
Essentials of Ultrasound Physics, Zagzebski J A, ISBN:
0815198523, 1996, 220 pp, Mosby.
Physical Principles of Medical Ultrasonics, Hill CR,
Bamber JC, ter Haar GR, ISBN: 0471970026,
2002, 528 pp, Wiley, New York.
346
Physics and Instrumentation of Diagnostic Medical

Ultrasound, Fish P, ISBN: 0471958956, 2005, 250
pp, Wiley, New York.
Principles and Applications of Ultrasound, Langton
CM, ISBN: 0750308052, 2005, 252 pp, IOPP, Bristol
(UK).
The Physics of Clinical MR Taught Through Images,
Runge VM, Nitz WR, Schmeets SH, Faulkner WH,
Desai NK, ISBN: 1588903222, 2004, 221 pp, Thieme
Med. Pub.
Ultrasound in Medicine, Duck FA, Baker AC, Starritt
HC, editors, ISBN: 0750305932, 1998, 314 pp, IOPP,
Bristol (UK).
Ultrasound Physics Mock Exam, Owen CA, Zagzebski
JA, ISBN: 0941022633, 2004, Davies, Inc.
Topical
AAPM Report No. 8, Pulse Echo Ultrasound Imaging
Systems: Performance Tests and Criteria, General
Medical Physics Committee Ultrasound Task Group;
AAPM Report No. 65, Real-Time B-Mode Ultrasound Quality Control Test Procedures (Reprinted
from Medical Physics, Vol. 25, Issue 8),
Ultrasound Task Group 1; ISBN: 1-888340-22-3,
(WI).
Basic Doppler Physics, Smith H, Zagzebski J, ISBN:
Doppler Ultrasound: Physics, Instrumental, and Clinical Applications, 2nd ed., Evans DH, McDicken
WN, ISBN: 0471970018, 2000, 456 pp, Wiley, New
York.
IPEM Report No. 70, Testing of Doppler Ultrasound
Equipment, Hoskins PR, Sherriff SB, Evans JA,
ISBN: 0904181715, 1994, 136 pp, IPEM, York
(UK).
IPEM Report No. 71, Routine Quality Assurance of
Ultrasound Imaging Systems ISBN: 0904181820,
1995, 66 pp, IPEM, York (U.K.).
IPEM Report No. 84, Guidelines for the Testing and
Calibration of Physiotherapy Ultrasound Machines,
Pye S, Zequiri B, ISBN: 0904181987, 2001, 67 pp,
IPEM, York (UK).
Safety of Diagnostic Ultrasound, Barnett SB, Kossoff G,
editors, ISBN: 1850706468, 1997, 147 pp, Taylor &
Francis.
The Safe Use of Ultrasound in Medical Diagnosis, ter
Haar G, Duck FA, ISBN 0-905749-42-1, 2000, 120 pp,
British Medical Ultrasound Society and British Institute of Radiology, London (UK).
Three-Dimensional Ultrasound, Downey B, Pretorius DH, Fenster A, ISBN: 0-7817-1997-6, 1999,
272 pp, Lippincott Williams & Wilkins, Philadelphia.
LIGHT AND LASERS

General
AAPM Report No. 3, Optical Radiations in Medicine: A
Survey of Uses, Measurement and Sources, AAPM,
An Introduction to Biomedical Optics, Splinter R, ISBN:
Applied Laser Medicine, Breuer H, Krasner N, Okunata
T, Sliney D, Berlien H-P, Mu ller GJ, ISBN:
354067005X, 2004, 740 pp, Springer, New York.
Laser-Tissue Interactions: Fundamentals and Applications, Niemz MH, ISBN: 3540405534, 2003, 305 pp,
Springer, New York.
Light, Visible and Invisible, and Its Medical Applications, Newing A, ISBN: 1860941648, 1999, 212 pp,
World Scientific, River Edge (NJ).
Topical
AAPM Report No. 57, Recommended Nomenclature for
Physical Quantities in Medical Applications of Light,
General Medical Physics Committee Task Group 2;
AAPM Report No. 73, Medical Lasers: Quality Control,
Safety, Standards, and Regulations, Task Group 6,
IPEM Report No. 76, Ultraviolet and Blue-Light Phototherapy-Principles, Sources, Dosimetry and Safety,
Diffey B, Hart G, ISBN: 0904181863, Published:
1997, 56 pp, IPEM, York (U.K.).
Laser Safety, Henderson R, ISBN: 0750308591, 2003,
480 pp, IOPP, Bristol (U.K.).
Laser Systems for Photobiology and Photomedicine,
Chester AN, Martellucci S, Scheggi AM, ISBN:
0306438860, 1991, 311 pp, Plenum, New York.
Ultraviolet Radiation in Medicine, Diffey BL, ISBN:
0852745354, 1982, 172 pp, IOPP, Bristol (U.K.).
RADIATION PROTECTION
General
Atoms, Radiation, and Radiation Protection, 2nd ed.,
Turner JE, ISBN: 0-471-59581-0, 1995, 576 pp,
Wiley, New York.
Basic Health Physics: Problems and Solutions,
Bevelacqua J, ISBN: 0471297119, 1999, 559 pp,
Wiley, New York.
Basic Radiation Protection Technology (2nd edition),
Gollnick DA, ISBN: 0916339033, 1988, Pacific Radiation Corp.
Contemporary Health Physics: Problems and Solutions,
Bevelacqua J, ISBN: 0471018015, 1995, 456 pp,
Wiley, New York.
CRC Handbook of Management of Radiation Protection

Programs, 2nd ed., Miller KL, ISBN: 0849337704,
1992, 496 pp, CRC Press, Boca Raton (FL).
Exposure Criteria for Medical Diagnostic Ultrasound:
2. Criteria Based on All Known Mechanisms
(NCRP Report, No. 140), ISBN: 0929600738, 2003,
NCRP.
Handbook of Health Physics and Radiological Health,
Shleien B, Slaback LA Jr, Birky B, ISBN:
0683183346, 1997, 700 pp, Lippincott.
Introduction to Health Physics, 3rd ed., Cember H,
ISBN: 00-71054618, 1996, 731 pp, McGraw-Hill,
New York.
Medical Effects of Ionizing Radiation, 2nd ed., Mettler
FA Jr, Upton AC, ISBN: 0721666469, 1995, 440 pp,
Elsevier.
Physics for Radiation Protection, Martin JE, ISBN:
0-471-35373-6, 2000, 713 pp, Wiley, New York.
Practical Radiation Protection and Applied Radiobiology, 2nd ed., Dowd SB, Tilson ER, editors,
ISBN: 0721675239, 1999, 368 pp, Saunders, New
York.
Practical Radiation Protection in Healthcare, Martin
CJ, Sutton DG, editors, ISBN: 0192630822, 2002,
440 pp, Oxford University Press, New York.
Radiation Protection, Seeram E, Travis E, ISBN: 0-39755032-4, 1996, 320 pp, Lippincott.
Radiation Protection, Kathren RL, ISBN: 0852745540,
1985, 212 pp, IOPP, Bristol (UK).
Radiation Protection, Hallenbeck WH, ISBN:
0873719964, 1994, 288 pp, CRC Press, Boca Raton
(FL).
Radiation Protection: A Guide for Scientists and Physicians, Shapiro J, ISBN: 0674745868, 1990, 494 pp,
Harvard University Press, Cambridge (MA).
Radiofrequency Radiation Standards: Biological
Effects, Dosimetry, Epidemiology, and Public Health
Policy, Klauenberg BJ, Grandolfo M, Erwin DN,
ISBN: 0306449196, 1995, 476 pp, Kluwer, Norwell
(MA).
Topical
AAPM Proceedings No. 4, Radiotherapy Safety,
Thomadsen B, editor, ISBN: 0-88318-443-5, Published: 1984, 169 pp, Medical Physics Publishing.
Madison (WI).
AAPM Report No. 19, Neutron Measurements Around
High Energy X-Ray Radiotherapy Machines, Radiation Therapy Committee Task Group 27, ISBN:
0-88318-518-0, 1986, 34 pp, Medical Physics Publishing. Madison (WI).
AAPM Report No. 25, Protocols for the Radiation Safety
Surveys of Diagnostic Radiological Equipment, Diagnostic X-Ray Imaging Committee Task Group 1;
Publishing. Madison (WI).
AAPM Report No. 35, Recommendations on Performance Characteristics of Diagnostic Exposure
347
Meters (Reprinted from Medical Physics, Vol. 19,

Issue 1), Diagnostic X-Ray Imaging Committee Task
Group 6; ISBN: 1-56396-029-X, 1992, 11 pp, Medical
Physics Publishing. Madison (WI).
Exposure of the Pregnant Patient to Diagnostic Radiations, Wagner L, et al., ISBN: 0-944838-72-3, 1997,
259 pp, Medical Physics Publishing, Madison
(WI).
How Clean is Clean, How Safe is Safe? Eisenbud M,
IPEM Report No. 50, Chernobyl: Response of Medical
Physics Departments in the UK, Haywood JK,
editor, ISBN: 0904181456, 1986, 99 pp, IPEM, York
(UK).
IPEM Report No. 63, Radiation Protection in Nuclear
Medicine & Pathology Goldstone KE, Jackson PC,
Myers MJ, Simpson A E, editors, ISBN: 0904181626,
1991, 190 pp, IPEM, York (UK).
IPEM Report No. 69, Recommendations for the Presentation of Type Test Data for Radiation Protection Instruments in Hospitals, The Radiation
Protection Instrument Calibration Working Party
of the IPEM, ISBN: 0904181707, 1994, 12 pp, IPEM,
York (UK).
IPEM Report No. 72, Safety in Diagnostic Radiology,
ISBN: 904181812, 1995, 119 pp, IPEM, York (UK).
IPEM Report No. 75, The Design of Radiotherapy Treatment Room Facilities, Stedeford B, Morgan HM,
Mayles WPM, ISBN: 1903613855, 1997, 161 pp,
IPEM, York (UK).
IPEM Report No. 82, Cost-Effective Methods of Patient
Dose Reduction in Diagnostic Radiology, ISBN:
0904181944, 2001, 50 pp, IPEM, York (UK).
IPEM Report No. 88, Guidance on the Establishment
and Use of Diagnostic Reference Levels for Medical
X-Ray Examinations, ISBN: 1903613205, 2004, 44
pp, IPEM, York (UK).
Management and Administration of Radiation Safety
Programs (HPS 1998 Summer School), Charles
Roessler, ISBN: 0-944838-01-4, 1998, 603 pp, Medical Physics Publishing, Madison (WI).
Public Protection from Nuclear, Chemical, and Biological Terrorism, Brodsky A, Johnson RH, Goans RE,
editors, ISBN: 1-930524-23-4, Published: July 2004,
Radiation Injuries, Gooden D, ISBN: 33333, 1991, 239
Radiation Protection Dosimetry: A Radical Reappraisal,
Simmons J, Watt D, ISBN: 0-944838-87-1, 1999, 160
Radiation Protection in Medical Radiography,
Statkiewicz Sherer MA, Visconti PJ, Ritenour RE,
ISBN: 0323014526, 2002, 336 pp, Mosby.
Radiation Safety and ALARA Considerations for the
21st Century, HPS Midyear Symposium, 2001,
348
Shielding Techniques for Radiation Oncology Facilities,

2nd ed., McGinley PH, ISBN: 1-930524-07-2, 2002,
184 pp, Medical Physics Publishing. Madison (WI).
Subject Dose in Radiological Imaging, Ng K-H, Bradley
DA, Warren-Forward HM, ISBN: 0-444-82989-x,
1998, Elsevier.
The Invisible Passenger, Radiation Risks for People
Who Fly, Barish RJ, ISBN: 188352606X, 1999, 119
pp, Advanced Medical Publishing, Madison (WI).
University Health Physics, Belanger R, Papin PJ, editors, ISBN: 1-930524-15-3, 2003, 408 pp, Medical
Physics Publishing. Madison (WI).
RADIATION MEASUREMENTS
General
Applications of New Technology: External Dosimetry,
Higginbotham JF, ISBN: 0-944838-68-5, 1996, 464
Fundamentals of Radiation Dosimetry, Green S, ISBN:
075030913x, 2005, 275 pp, IOPP, Bristol (UK).
Medical Radiation Detectors: Fundamental and Applied
Aspects, Kember NF, ISBN: 0750303190, 1994, 236
pp, IOPP, Bristol (UK).
Radiation Detection and Measurement, 3rd ed., Knoll
GF, ISBN: 0-471-07338-5, 1999, 816 pp, Wiley, New
York.
Radiation Dosimetry: Physical and Biological Aspects,
Orton CG, ISBN: 0306420562, 1986, 344 pp, Plenum,
New York.
Radiation Instruments, Cember H, ISBN: 1-93052403-X, 2001, 472 pp, Medical Physics Publishing,
Madison (WI).
Topical
A Procedural Guide to Film Dosimetry, Yeo IJ, Kim JO,
ISBN: 1-930524-19-6, 2004, 65pp, Medical Physics
AAPM Proceedings No. 11, Kilovoltage X-Ray Dosimetry for Radiotherapy and Radiobiology, Ma C-M,
Seuntjens J, editors, ISBN: 1-888340-16-9, 1998,
(WI).
AAPM Proceedings No. 13, Recent Developments in
Accurate Radiation Dosimetry, Seuntjens JP, Mobit
PN, editors, ISBN: 1-930524-12-9, 2002, 353 pp,
AAPM Report No. 12, Evaluation of Radiation Exposure
Levels in Cine Cardiac Catheterization Laboratories,
Diagnostic Radiology Committee Cine Task Force;
AAPM Report No. 63, Radiochromic Film Dosimetry
(Reprinted from Medical Physics, Vol. 25, Issue
11), Radiation Therapy Committee Task Group 55;
Instrumentation, Measurements and Electronic Dosimetry, (HPS Midyear 2000) ISBN: 0-944838-93-6,
(WI).
Internal Radiation Dosimetry (1994 HPS Summer
School), Raabe O, ISBN: 0-944838-47-2, 1994,
(WI).
Microdosimetry and Its Applications, Rossi HH, Zaider
M, ISBN: 3540585419, 1996, 321 pp, Springer, New
York.
Practical Applications of Internal Dosimetry, Bolch WE,
editor, ISBN: 1-930524-09-9, 2002, 480 pp, Medical
RADIATION BIOLOGY
General
An Introduction to Radiobiology, Nias AHW, ISBN:
0471975907, 1998, 400 pp, Wiley, New York.
Basic Clinical Radiobiology, Steel GG, ISBN:
0340807830, 2002, 266 pp, Edward Arnold, London.
Biological Risks of Medical Irradiations, Fullerton G,
ISBN: 0883182793, 1987, 335 pp, American Institute
of Physics, New York.
Effects of Atomic Radiation, Schull WJ, ISBN:
0471125245, 1995, 97 pp, Wiley, New York.
Handbook of Radiobiology, Prasad KN, ISBN:
0849325013, 1995, 352 pp, CRC Press, Boca Raton
(FL).
Primer of Medical Radiobiology, Travis E, ISBN:
0815188374, 1989, 302 pp, Mosby, Philadelphia.
Radiobiology for the Radiologist, 5th ed., Hall E J, ISBN:
0-7817-2649-2, 2000, 608 pp, Lippincott.
Topical
A Compilation of Radiobiology Practice Examinations
for Residents in Diagnostic Radiology and Radiation
Oncology, Chapman JD, Shahabi S, Chapman BA,
ISBN: 1 883526 09 4, 2000, 163 pp, Advanced Medical
AAPM Proceedings No. 7, Prediction of Response
in Radiation Therapy: Analytical Models and
Modelling, Paliwal B et al., ISBN: 0-883186-24-1,
(WI).
AAPM Proceedings No. 9, Prediction of Response in
Radiation Therapy: Radiosensitivity, Repopulation,
Paliwal BR, Herbert D, Fowler JF, Kinsella TJ,
AAPM Proceedings No. 10, Volume & Kinetics in
Tumor Control & Normal Tissue Complications:
5th International Conference on Dose, Time,
and Fractionation in Radiation Oncology, Paliwal
BR, Herbert D, editors, ISBN: 1-888340-11-8, 1998,
(WI).
AAPM Report No. 18, A Primer on Low-Level Ionizing

Radiation and Its Biological Effects, Biological
Effects Committee; ISBN: 0-88318-514-1, 1986,
(WI).
AAPM Report No. 43, Quality Assessment and Improvement of Dose Response Models: Some Effects of Study
Weaknesses on Study Findings, Biological Effects
Committee Task Group 1; ISBN: 0-944838-45-6,
(WI).
Applied Radiobiology and Bioeffect Planning, Wigg D,
ISBN: 1-930524-05-6, 2001, 486pp, Medical Physics
Biological Models and Their Applications in
Radiation Oncology, Olsen KJ, ISBN: 10883526-035, 1994, 61 pp, Advanced Medical Publishing,
Madison (WI).
BJR Supplement 26: Chronic Irradiation: Tolerance and
Failure in Complex Biological Systems, 2002, The
British Institute of Radiology.
Health Effects of Exposure to Low-level Ionizing Radiation, Hendee WR, Edwards FM, editors, ISBN:
Health Effects of Exposure to Low Levels of Ionizing
Radiation: BEIR V, National Research Council,
ISBN: 0309039959, 1989, 436 pp, National Academies Press.
Physics and Radiobiology of Nuclear Medicine, Gopal
Saha, 2nd ed., ISBN: 0387950214, 2003, 253 pp,
Springer, New York.
Prediction of Tumor Treatment Response, Chapman
DJ, Peters LJ, Withers RH, ISBN: 0080346898,
1989, 336 pp, Elsevier.
Radiation Oncology Radiobiological and Physiological
Perspectives, Awwad H K, ISBN: 0792307836, 1990,
688 pp, Kluwer, Norwell (MA).
The Radiation Biology of the Vascular Endothelium,
Rubin DB, ISBN: 0849348404, 1997, 272 pp, CRC
Press, Boca Raton (FL).
RADIOLOGICAL PHYSICS FOR RADIOLOGICAL

TECHNOLOGISTS
349
Topical
Magnetic Resonance Imaging: Physical and Biological
Principles, Bushong SC, ISBN: 0323014852, 2003,
528 pp, Mosby, Philadelphia.
MRI for Technologists, Woodward P, ISBN:
0071353186, 2000, 408 pp, McGraw-Hill, New York.
Principles and Practice of Radiation Therapy,
Washington CM, Leaver D, ISBN: 0323017487,
2004, 992 pp, Elsevier, New York.
Rad Techs Guide to Mammography: Physics, Instrumentation, and Quality Control, Jacobson DR,
Seeram E, ISBN: 0632044993, 2001, 120 pp, Blackwell.
Radiation Protection: Essentials of Medical Imaging
Series, Stewart C. Bushong, ISBN: 0070120137,
1998, 288 pp, McGraw-Hill, New York.
The Basic Physics of Radiation Therapy, Selman J,
ISBN: 0398056854, 1990, 749 pp, Charles C. Thomas.
The Fundamentals of Imaging Physics and Radiobiology for the Radiologic Technologist, Selman J, ISBN:
0398069875, 2000, 484 pp, Charles C. Thomas.
MATHEMATICS AND STATISTICS
AAPM Monograph No. 13, Multiple Regression Analysis: Applications in the Health Sciences, Herbert DE,
Meyers R H, editors, ISBN: 0-88318-490-7, 1984, 598
Chaos and the Changing Nature of Science and Medicine: An Introduction, Herbert D, editor, ISBN:
1-56396-442-2, 1995, American Institute of Physics,
New York.
Mathematical and Computer Modeling of Physiological
Systems, Rideout V, ISBN: 0-13-563354-0, 1991, 155
pp, Prentice Hall, New York.
COMPUTERS
General
AAPM Monograph No. 17, Computers in Medical Physics (1988 Summer School), Benedetto A R, Huang
HK, Ragan DP, editors, ISBN: 0-88318-802-3, 1988,
General
Introduction to Radiologic Sciences and Patient Care,
3rd ed., Adler AM, Carlton RR, ISBN 0721697828,
2003, 520 pp, Saunders, Philadelphia.
Radiologic Physics and Radiographic Imaging,
Bushong SC, ISBN: 0323032648, 2004, Mosby,
Philadelphia.
Radiologic Science for Technologists: Physics, Biology
and Protection, Bushong SC, ISBN: 0323025552,
2004, 704 pp, Mosby.
The Fundamentals of X-Ray and Radium Physics,
Joseph Selman, ISBN: 0398058709, 1994, 637 pp,
Charles C. Thomas.
Topical
AAPM Report No. 10, A Standard Format for Digital
Image Exchange, Task Force on Digital Image Data
Exchange of the Science Council; ISBN: 0-88318408-7, 1982, 11 pp, Medical Physics Publishing,
Madison (WI).
AAPM Report No. 30, E-Mail and Academic Computer
Networks, Computer Committee Task Group 1,
PACS: Basic Principles and Applications, Huang HK,
350
PUBLIC EDUCATION
AAPM Report No. 53, Radiation Information for Hospital Personnel, Radiation Safety Committee; ISBN:
Cancer Patients Guide to Radiation Therapy, Steeves
R, ISBN: 0-944838-26-X, 1991, 87 pp, Medical Physics Publishing, Madison (WI).
Radiation and Health, Thormod Henrikson, H. David
Maillie, David H. Maillie, ISBN: 0415271622, 2002,
240 pp, Taylor & Francis.
MEDICAL PHYSICS EDUCATIONAL AND

PROFESSIONAL ISSUES
AAPM Monograph No. 27, Accreditation Programs and
the Medical Physicist: 2001 AAPM Summer School
Proceedings, Dixon R, Butler P, Sobol W, editors,
ISBN: 1-930524-04-8, 2001, 364 pp. Medical Physics
AAPM Report No. 33, Staffing Levels and Responsibilities of Physicists in Diagnostic Radiology, Diagnostic X-Ray Imaging Committee Task Group 5; ISBN:
AAPM Report No. 36, Essentials and Guidelines for
Hospital Based Medical Physics Residency Training
Programs, Presidential Ad Hoc Committee on Clinical Training of Radiological Physicists; ISBN:
AAPM Report No. 38, The Role of a Physicist in Radiation Oncology, Professional Information and Clinical
Relations Committee Task Group 1; ISBN: 1-56396229-2, 1993, 12 pp, Medical Physics Publishing,
Madison (WI).
AAPM Report No. 42, The Role of the Clinical Medical
Physicist in Diagnostic Radiology, Professional Information and Clinical Relations Committee; ISBN:
AAPM Report No. 64, A Guide to the Teaching of
Clinical Radiological Physics to Residents in Diagnostic and Therapeutic Radiology, Committee on the
Training of Radiologists; ISBN: 0-944838-09-X,
(WI).
AAPM Report No. 79, Academic Program Recommendations for Graduate Degrees in Medical Physics,
Education and Training of Medical Physicists Committee, ISBN: 1-888340-39-8, 2002, 72 pp, Medical
AAPM Report No. 80, The Solo Practice of Medical
Physics in Radiation Oncology, Professional Information and Clinical Relations Committee Task
Group 11, ISBN: 1-888340-41-X, 2003, 18 pp, Medical
Current Regulatory Issues in Medical Physics, Martin

M, Smathers J, ISBN: 0-944838-29-4, 1992, 458 pp,
Medical Physicists and Malpractice, Shalek R, Gooden
D, ISBN: 0-944838-64-2, 1996, 140 pp, Medical Physics Publishing, Madison (WI).
RADIATION PHYSICS
AAPM Proceedings No. 8, Biophysical Aspects of Auger
Processes, Howell RW, Narra V, Sastri K, Rap D,
AAPM Report No. 37, Auger Electron Dosimetry (Reprinted from Medical Physics, Vol. 19, Issue 6),
Nuclear Medicine Committee Task Group 6; ISBN:
AAPM Report No. 49, Dosimetry of Auger-ElectronEmitting Radionuclides (Reprinted from Medical
Physics, Vol. 21, Issue 12), Nuclear Medicine Committee Task Group 6; ISBN: 1-56396-451-1, 1994, 15
Bluebells and Nuclear Energy, Reynolds A, ISBN: 0944838-63-4, 1996, 302 pp, Medical Physics Publishing, Madison (WI).
Chernobyl Record: The Definitive History of the Chernobyl Catastrophe, Mould RF, 075030670x, 2000,
My Life with Radiation, Ralph Lapp, ISBN: 0-94483852-9, 1995, 168 pp, Medical Physics Publishing,
Madison (WI).
Radiological Physicists, del Regato JA, ISBN: 0-88318469-9, 1985, 188 pp, Medical Physics Publishing,
Madison (WI).
Understanding Radiation, Wahlstrom B, ISBN:
Medical Physics Online Resources
Electronic Medical Physics World, International
Organization for Medical Physics (http://www.
medphysics.wisc.edu/empw).
Medical Physics World, International Organization for
Medical Physics (http://www.iomp.org); also available in hardcopy.
Global Online Medical Physics Book (http://www.iomp.
org).
Organizations That Publish in Medical Physics
(AAPM), One Physics Ellipse, College Park, MD
20740 (http://www.aapm.org).
American Institute of Physics, 2 Huntington Quadrangle, Melville, NY 11747-4502 (http://aip.org).
American Institute of Ultrasound in Medicine, 14750
Sweitzer Lane, Suite 100, Laurel, MD 20707 (http://
www.aium.org).
American Roentgen Ray Society, 44211 Slatestone Court,

Leesburg, VA 20176-5109 (http://www.arrs.org).
American Society for Therapeutic Radiology and Oncology, 12500 Fair Lakes Circle, Suite 375, Fairfax, VA
22033-3882 (http://www.astro.org).
British Institute of Radiology, 36 Portland Place,
London, W1B 1AT, (http://www.bir.org.uk).
Health Physics Society, 1313 Dolley Madison Boulevard,
Suite 402, McLean, Virginia 22101 (http://hps.org).
Institute of Physics and Engineering in Medicine
(IPEM), Fairmount House, 230 Tadcaster Road,
York, YO24 1ES, UK (http://www.ipem.ac.uk).
International Atomic Energy Agency, P.O. Box 100,
Wagramer Strasse 5, A-1400 Vienna, Austria
(http://www.iaea.org).
International Commission on Radiation Units and Measurements, Inc., ICRU, 7910 Woodmont Avenue,
Suite 400, Bethesda, MD 20814-3095, USA (http://
www.icru.org).
International Commission on Radiological Protection,
SE-171 16 Stockholm Sweden (Elsevier).
International Society for Magnetic Resonance in Medicine, 2118 Milvia Street, Suite 201, Berkeley, CA
94704, USA (http://www.ismrm.org).
International Society for Optical Engineering (SPIE),
PO Box 10, Bellingham WA 98227-0010 USA (http://
spie.org).
National Council on Radiation Protection and Measurements, 7910 Woodmont Avenue, Suite 400, Bethesda,
MD 20814-3095 (http://www.ncrponline. org).
Radiation Research Society, 10105 Cottesmore Court,
Great Falls, VA 22066 (http://www.radres.org).
Radiological Society of North America, Inc., 820 Jorie
Boulevard, Oak Brook, IL 60523-2251 (http://rsna.org).
Society of Nuclear Medicine, 1850 Samuel Morse Dr.
Reston, VA 20190 (http://snm.org).
See also BIOMEDICAL ENGINEERING EDUCATION; MEDICAL EDUCATION,
COMPUTERS IN; MEDICAL ENGINEERING SOCIETIES AND ORGANIZATIONS.

YESHWANTH SRINIVASAN
BRIAN NUTTER
Texas Tech University
Lubbock, Texas
INTRODUCTION
Computers play a vital role in almost every aspect of
modern medical science. If we begin with a primitive
definition of a computer, as a system with a processor unit
capable of performing arithmetic and logical operations
and a memory unit to store programs and data, then almost
every sophisticated piece of modern medical equipment
comes with a computer. Most standard medical procedures
[e.g., magnetic resonance imaging (MRI), computed tomo-
351
graphy (CT), positron emission tomography (PET), ultrasonography] use computers to record and to process data.
Computer-based medical records would seem a natural
choice for documentation of the recorded data. Moreover,
the association of this data with computers presents
numerous advantages including easy and random access
to the data by multiple distributed users, globally searchable databases, and automated batch processing. However,
the use of computer-based medical records is not as prevalent in todays medical community as one would expect.
In this article, we explain the reasons for this anomaly, and
we present methods to circumvent the problems associated
with computer-based medical records and to allow medical
practitioners to achieve the real potential of computerbased systems.
PAPER-BASED MEDICAL RECORD SYSTEMS
Medical records chronicle the transactions between healthcare professionals and their patients. They record every
detail of a patients visit to a hospital, clinic, or office from
the time of the patients arrival to the time the patient is
discharged. Details recorded include the condition of the
patient at the time of examination, procedures and medications administered to the patient, symptoms and observations, diagnoses and test results, together with the exact
time of each relevant activity.
Traditionally, medical records have consisted of handwritten documents called charts. This remains the most
prevalent form of recording patient information, especially
in small hospitals and clinics. These charts typically record
every piece of information that is judged potentially relevant to a patients case. When a patient visits a hospital or a
primary care physician for the first time, a new chart is
made, and the patient name and contact information are
recorded on it. When the patient is discharged from the
hospital, a discharge note is made on the chart. During
subsequent visits, new charts are added. All patient charts
are maintained in a labeled file bearing the patients name
and other information to enable easy identification and
retrieval.
However, paper-based records exhibit shortcomings in
six important areas that are necessary to ensure adequate
long-term care to patients and easy preservation of treatments administered.
1. Accessibility: Accessibility refers to the immediate
availability of patient-related information at all
times to all parties with a need to know. It is important to review a patients medical history before
commencing any new treatment. If a patient decides
to visit a specialist, emergency room, clinic, or hospital other than the primary care physician, a
request may be made for the patients medical history
to be faxed or couriered to the second facility. This
information transfer may take between hours to days
to complete, depending on the distance and arrangements between the two locations, available modes of
communication, availability of authorized personnel
to promptly respond to the request, and so on. During
352
2.
3.
4.
5.
emergencies, a delay of just a few minutes in administering the correct treatment to a patient could have
very serious ramifications. A situation in which
adverse reactions to particular medications known
by a primary care physician simply are not made
available in a timely fashion to emergency room
personnel is all easy to imagine. The delay in disseminating accurate detailed background information is perhaps the biggest drawback of paper-based
medical record systems.
Legibility: Paper-based records are generally handwritten documents, and one persons handwriting is
often not readily understood by another. The fact that
entries may be made by many different people makes
understanding a chart even more difficult. Furthermore, paper-based records are subject to wear and
tear, accidental immersion in water or other liquids
due to leaks, misfortune or negligence, smudging of
ink, and so on. These factors lead to reduced legibility
of the entries on the charts and consequently make
interpreting the charts more subjective.
Bulkiness: Paper-based records for a single patient
initially consist of just a few charts, but they can
quickly become a thick file with subsequent visits.
Assuming that a small clinic sees 10 new patients
daily, that the physicians in the clinic take rotating
holidays and that it is not a leap year, 3650 new files
will be added over a period of 1 year. The files of many
of the existing patients also become bulkier and subsequently occupy more space. All of this paper forces
the addition of storage space in the clinic for these new
records, even though the space could often be better
utilized for clinical purposes through the addition of
much more profitable and useful treatment or diagnostic capabilities.
Data Dimensionality: Data dimensionality refers to
the ability to interpret the available information in
multiple, complementary fashions. The data contained in paper-based records simply provide information about the medical history of a particular patient.
In order to obtain trends, distributions or statistical
patterns about a particular patient or even groups of
people in a geographical area, the required information has to be manually compiled from each record.
This can be extremely laborious and time consuming.
Furthermore, even a simple text search for a particular word or phrase can take impractically long if the
search must be conducted visually through paperbased handwritten records.
Integrity: A patient will have multiple medical files,
maintained simultaneously at every hospital or clinic
ever visited, yet none of these records contain the
complete medical history of the patient. Usually, the
patients primary care physician is the one who has the
records that will provide the most accurate long-term
information, and during times of emergency the primary care physician will frequently be the first one
contacted for patient records. However, medical facilities rarely have in place any mechanism to update the
primary care physician. If, for example, during a visit
to a specialist it was found that a patient was allergic to

a particular class of medications, the primary care
physicians records will often not be updated to show
this information. Even if the patient records from
multiple sources were summoned, there will still
remain a very realistic chance of missing one or more
sources, leaving the completeness of paper-based
records questionable. Paper-based systems also risk
inadvertent loss or misfile of complete pages in copying, transporting, photocopying, faxing, and handling
of records.
6. Replication: Paper-based medical records are prone to
destruction by natural disasters, fires, leaks, and pests
in storage areas. Because paper-based records are
bulky, it is both expensive and tedious to replicate
them and store back-up copies. Hence, if they are
destroyed, patient medical histories can be completely
lost.
Several research papers have been devoted to explaining the shortcomings of paper-based medical records (1,2).
Nevertheless, paper-based records are still the most widely
used method for maintaining patient information. The
process of recording information in charts is a natural
process of making notes, and it requires little specialized
training. Once recorded, the information is relatively permanent and cannot be distorted easily. Furthermore, the
chart often bears the signatures of both the patient and the
appropriate medical personnel who treated the patient,
which may be an important legal requirement. Such legal
issues are a significant reason that many organizations
resist wholesale changes to their paper-based record
systems.
COMPUTER-BASED MEDICAL RECORD SYSTEMS
Paper-based records were the only method of storing
patient medical information in use until 25 years ago.
Since that time, computer technology has experienced
exponential growth, creating endless possibilities for developments within allied fields. The fields of medicine in
general and medical records in particular have been no
exception. Computer-based records were introduced to
overcome the inherent limitations and problems of
paper-based records and to alter patient medical record
keeping in order to incorporate advancements in computer
technology. The computer-based medical record is commonly given several names within differing environments,
including Computer-based Patient Record (CPR), Electronic Health Record (EHR), and Electronic Medical Record
(EMR). Throughout this article, we will use the name
computer-based patient record and the abbreviation CPR
to refer to computer-based medical records. The CPRs are
fundamental to the role of computers in medical records,
and a significant portion of this article deals with them.
A CPR is an electronic patient record that resides in a
computer-based information system. Such systems are
often specifically designed to augment medical practitioners by providing accessibility to complete and accurate
long-term data, alerts, reminders, clinical decision support
systems, links to medical knowledge databases, and other

aids (3,4). The CPRs are essentially digital equivalents of
paper-based records. Because this patient information is
stored digitally, the enormous information processing and
networking capabilities of computers can be utilized to
drastically increase the efficacy of patient treatment. They
easily overcome the shortcomings of paper-based records
on the six key issues outlined in the previous section, and
they provide several other fascinating features exclusive to
them.
1. Accessibility: With the advent of the Internet and the
World Wide Web, information on virtually any topic
can be quickly and easily obtained. CPR databases
can be stored in networked servers, which can be
searched from any properly networked location. With
properly designed databases and network systems,
the complete medical history of a patient can be
retrieved in a matter of seconds. This helps in ensuring the most prompt and accurate treatment possible, which is one of the fundamental reasons for
maintaining a medical record.
2. Legibility: The information recorded on CPRs is presented using digitally reproduced fonts, which are
then independent of the handwriting of the person
entering the data into the system. Anytime patient
information is required, it can be readily visualized
on a viewing device, such as a CRT monitor or PDA,
which does not introduce any wear and tear in the
record itself. If a paper copy of a patient record is
required for a particular activity, a new printout can
be generated every time. Thus there is no subjective
element in understanding the information.
3. Bulkiness: CPRs can be stored on a variety of media,
including Compact Disc (CD), Digital Versatile Disc
(DVD), and Hard Disk Drive (HDD). A single CD,
11.4 cm in diameter and weighing 15 g, can store
700 megabytes (MB) of data. This translates into
>100,000 pages of raw text or two hundred 500-page
books, which would occupy >10 m of shelf space. The
DVDs have >10 times the density of CDs, and HDDs
offer >100 times the information content of a CD.
This means that with proper database design, all of
the medical records of all of 1 years new patients at
the aforementioned small clinic can now be stored on
a single CD, and several years of patient data for all
ongoing patient activities can be stored on just one
DVD.
4. Data Dimensionality: Data on CPRs can be exploited
and interpreted in many more ways than can a
simple paper-based record. Entire databases can be
subjected to automated search, and specific information from a single record or a collection of records can
be retrieved in a matter of seconds. Trends and
patterns in disease occurrence and treatment administration and effectiveness can readily be extracted
and processed for limited or wide-ranging populations. Database Management System (DBMS) software can be used to plot the available information on
graphs and pie charts, which greatly simplify the
353
contents of medical records into a form that can be

interpreted by laymen. All these operations can be
done automatically using state-of-the-art software
and require little, if any, human intervention. Techniques in data mining, for example, allow automated
determination and analysis of data correlations
among very large numbers of variables.
5. Integrity: CPRs are generally stored on local systems
within hospitals, clinics, and physician offices, which
means that medical records of a particular patient
could be distributed across many systems throughout
the world. Using modern networking techniques,
these systems can be easily yet securely networked,
and widely dispersed CPRs can be located and made
available for inclusion in patient diagnosis and treatment. It would also be practical to develop a system
in which all patient records are stored in a central
repository, and any time new records of the
patient are created or updated, they can be added
to this central repository. Using either of these methods in a well-designed system, the complete medical
history of a patient can be very rapidly retrieved from
any of the authorized locations connected to the
network.
6. Replication: Due to the high densities for computerized storage of data, CPRs occupy very little space
when compared to paper-based systems. It is very
easy and inexpensive to create back-up copies and to
store these copies in geographically diverse locations.
In a well-designed computer-based system with automated, networked backup, the complete medical history of every patient can still be retrieved even if a
record storage facility is completely destroyed.
Exclusive Features of CPR
With considerable manual effort, the performance of paperbased records can be made comparable to that of CPRs
within the six major requirements of a medical record
outlined above. The efficiency advantages of computerbased medical record systems will nevertheless offer significant cost advantages over paper-based systems in
achieving comparable performance levels. Significant
improvements in patient treatment can result when CPRs
are incorporated into expert systems and DBMS frameworks that engender an entirely new set of features for
medical records.
1. Content-Based Data Retrieval: If a large hospital
system is informed that a popular drug has been
determined to cause serious side effects and that
the manufacturer has consequently issued an
immediate recall, the hospital system may hesitate
to issue a broad public warning, because that action
would lead to unnecessary panic and to patients
flooding the system seeking confirmation on their
prescriptions, even when their medications are completely unrelated to the recall. If the hospital maintained only paper-based records, it would take weeks
to pore through patient charts to obtain the names
354
2.
3.
4.
5.
and locations of the patients to whom the drug was

prescribed. Such a situation presents a perfect case
for content-based data retrieval. With properly constructed CPRs, the same information can be much
more rapidly retrieved through querying the database to fetch all patient records with the name of the
recalled drug in the Drugs Prescribed field. Then,
only the patients who actually have been prescribed
the recalled medication can be individually contacted
with explicit instructions by a qualified staff member.
This data search technique is known as ContentBased Data Retrieval (CBDR).
Automatic Scheduling: With CPR-based systems,
routine periodic check-up or follow-up visits can be
automatically scheduled, and e-mail reminders or
automated telephone recordings can be sent directly
to the patients. This automation will relieve healthcare professionals of a significant administrative
burden.
Knowledge-Based Systems: Every physician attempts
to solve a particular condition to the best of his
limited knowledge (5). It is reported that there are
> 5,000,000 medical facts, which are constantly
being appended, updated and questioned by over
30,000 medical journal publications each year (6).
It is humanly impossible to remember all of these
facts and to reproduce them accurately as needed.
However, CPR databases all over the world can be
rapidly and effectively searched in order to find
information about patients who were previously
diagnosed with the same condition, how they were
treated and the end result. While the presiding
doctor still makes the decisions and retains responsibility, he can have experimental evidence and
experiential knowledge from other experts to assist
his analysis (7). Diagnostic decision support systems like Quick Medical Reference (QMR) and
Massachusetts General Hospitals DXplain provide
instant access to knowledge bases of diseases,
diagnoses, findings, disease associations and lab
information.
Expert Systems: Human errors in recording information can be greatly reduced by programming the
data-entry interface to reject anomalous values
and to restrict the input to a finite set of possible
choices. For example, a pulse rate of 300 beats min1
would be highly unlikely. Potentially dangerous spelling mistakes in generating a prescription can be
identified. Dates for follow-up activities can be
checked against reasonable parameters. A welldesigned interface would prompt a message to
recheck such entries.
In situ Data Access: A medical records system in
which a physician accesses a wireless PDA while
conversing with a patient could allow direct bidirectional communication between a physician and a
pharmacy, a laboratory, or a specialist. Collaborators
on a complex case can simultaneously access each
others progress, even when the collaborators are
geographically separated.
In 1991, the United States Institute Of Medicine (IOM)

conducted a study on the advantages and disadvantages of
CPRs and published its findings (3). The IOM concluded
that CPRs greatly enhance the health of citizens and greatly
reduce costs of care, and it called for widespread implementation of CPRs by 2001. However, 4 years past that deadline,
widespread implementation still remains only a concept.
Figures for CPR adoption rates in hospitals range from 3%
to 21%, while CPR adoption rates for physician offices range
from 20% to 25% (8). It is safe to say that these figures are
not representative of the popularity that one would reasonably expect from the benefits of CPRs. The following section
is devoted to explaining this anomaly.
ROADBLOCKS IN CPR IMPLEMENTATION
Recording and processing of information with CPRs
requires installation of computer systems and software,
and, in general, more training of healthcare professionals
is required to get them acquainted with CPR systems than
with paper-based record systems. Furthermore, realizing
the full potential of CPR systems requires not only that
patient records issued in the future should be CPRs but
also that previously taken paper-based records should be
converted into CPRs. Some of the important factors affecting this onerous large-scale conversion will now be discussed.
1. Cost: Expense is usually the major obstacle to the
full-fledged incorporation of CPRs. Depending on
whether the application is for the office of a single
physician or for a large hospital, the basic infrastructure needed to establish a CPR system could vary
from a single PC to a complete network of workstations and servers loaded with expensive network,
database and records management software. Both
hardware and software computer technologies
become outmoded very quickly, and these technologies require constant update, increasing maintenance and operation costs. There is also a
significant cost factor involved in hiring data-entry
operators to convert existing paper-based records to
CPRs and in training healthcare professionals to
then operate the new systems and software.
Although several studies have shown that CPRs
are effective in the long run (9,10), both hospitals
and physician offices remain apprehensive about
investing many thousands of dollars on a new technology while the tried and tested paper-based records
work reasonably effectively.
2. Abundance of Choices and Lack of Standards: There
are > 100 CPR management software systems available in the market today, and additional entries
arrive in the marketplace annually. Each software
approach will have its advantages and disadvantages, and it is never easy to decide which one is best
suited for a particular application. Furthermore,
different software interfaces use different templates
for the patient record, recording slightly different
patient data within very different structures. A
3.
4.
5.
6.
standardized CPR format independent of the specific

hardware and software running on a particular system is yet to be developed. This lack of standardization in CPR formats severely cripples cross-system
interaction and comprehensive integration of CPR
systems. The marketplace has, to date, failed to
produce a clear market leader or an industrial body
with the ability to enforce such standards. A customer contemplating a purchase and the subsequent
investments in training and maintenance may have
serious reservations concerning the ability to change
from the products of one manufacturer to another.
Confidentiality: The United States government
passed the Health Insurance Portability and Accountability Act (HIPAA) in 1996, protecting a patients
right to confidentiality and specifying the information on a medical record that can be shared and the
information that must be considered confidential. A
record can be made available only on a need to
know basis, wherein only personnel who absolutely
need to know the data are granted access. For example, physicians participating in the diagnosis and
treatment process of a specific patient are granted
complete access to the entire medical record, while
health agencies involved in conducting demographic
studies are granted access only to information that
does not and can not reveal the identity of the
patient.
Security: In the case of paper-based records, it is
relatively easy to control access and to ensure security by keeping those records under lock and key. The
security of CPRs will be constantly threatened by
hackers, trying to break into a system in order to
retrieve confidential patient information. Unless
properly protected, CPRs are also vulnerable to being
compromised during transmission across a network.
Unlike paper-based records, CPRs can be easily manipulated, intentionally or unintentionally, in fashions
that do not appear obvious. These vulnerabilities
represent an enormous security problem, and effective solutions are required before CPRs can be put to
widespread use. Furthermore, efforts to provide network security have significant costs, requiring manpower, equipment, and frequent software upgrades.
Apprehension Among Patients: Even in developed
countries like the United States, considerable uneasiness exists among patients about the use of computers to maintain patient medical records. While
some of this phenomenon can be ascribed to the
reluctance of the older generation in accepting new
technology, the most important reason for this apprehension is the lack of clear understanding by patients
about the benefits of CPRs. One of the easiest ways to
overcome this fear is for healthcare professionals to
directly reach out to the patients and explain the
long-term advantages of CPRs.
Legal Issues: The validity of CPRs in a court of law is
still questionable. The Millennium Digital Commerce Act, otherwise known as the Electronic Signature Act, was passed in 2000 (although the portion
355
relating to records retention was passed in 2001). The

act established that the electronic signature and
electronic records cannot be denied legal effect simply because they are electronic and are not signed in
ink on paper. However, legal implications in CPR
transactions extend far beyond this act. For example,
if it is found that the software system used to manage
CPRs does not comply with stipulated security
requirements, the validity of a CPR can be questioned in court, even if no indication of tampering
can be identified. The CPR systems are also prone to
computer viruses, worms and Trojans, which exploit
loopholes in the computer security defenses to gain
unauthorized access to sensitive data and/or to maliciously corrupt the information.
Despite significant progress toward overcoming these
hurdles, complete transition from paper-based records to
CPRs is a time-consuming and capital-intensive process,
and during the midst of this transition, the benefits are
unlikely to be immediately apparent. In the following
section, some of the options available to healthcare professionals in implementing a CPR system, and several technologies available to ensure secure storage and transmission
of confidential medical information are reviewed.
FEATURES OF CPR SYSTEMS
A complete CPR system can consist of individual workstations, data input devices (keyboards, mice, light pens,
scanners, cameras), output devices (printers, plotters, and
display monitors), Database Management System (DBMS)
software, text, image and video processing hardware and
software, and the networking infrastructure for intra- and
intersystem communication. The security and reliability
requirements of these systems should never be understated. With so many factors involved, there is always
room for improvement, and a universally perfect combination has yet to be implemented. Nevertheless, a wide
variety of CPR systems have been successfully implemented, tested and put to everyday use, with most such systems
working successfully in the environment for which they
were designed. We will now explore some of the general
features of CPR systems and their components, and we
will present advantages and disadvantages of the choices
available.
System Architecture
A fundamental decision that must be made before implementing a CPR system is the choice of the system architecture to be used. There are two broad choices, which we
now discuss.
1. Centralized Architecture: A block schematic of a centralized architecture as implemented in CPR systems
is shown in Fig. 1. A CPR system based on the
centralized architecture has a central server that
contains complete medical records of all patients in
the system. Any modification to a record must be
356
Figure 1. Centralized Architecture (Reprinted with permission

from (The Computer-Based Patient Record: An Essential
Technology for Health Care, Revised Edition) # (1997) by the
National Academy of Sciences, courtesy of the National
Academies Press, Washington, D.C.)
done to the CPR stored in this central repository to

have any lasting effect. Communication links may be
provided to other departments that will immediately
communicate any new information to the server,
thereby keeping the CPR on the central server fully
updated. The principal advantage of this architecture lies in the immediate availability of the complete
record for use elsewhere within the system. It is also
easier to manage and to maintain the resources of the
system, because they are located in one single location. However, a poorly designed system can be prone
to server overloading, leading to delayed responses.
Server or network failures can bring the system to a
complete standstill, and if the data on the server is
corrupted or destroyed, the medical record activity
subsequent to system backup may be completely lost.
These issues can be resolved, but the solutions can be
expensive.
2. Distributed Architecture: The alternative to the centralized approach is the distributed architecture. A
block schematic is shown in Fig. 2. In this approach,
the load of the server is distributed among smaller
subsystems, distributed through many departments.
No department contains the complete medical record
of all patients. The records are completed on
demand, that is, when a user at a system workstation must obtain complete information, the workstation sends requests to each individual subsystem,
receives data back from them, and arranges them
into a record. This system continues to function even
if one or more of the subsystems become inoperative.
However, the response time, defined as the time
elapsed between the user request to fetch a record
and the arrival of the complete record back to the
requesting workstation, might become significant if
even one of the subsystems is overloaded. The architecture also provides more opportunities for security
breaches than the centralized approach. There may
Figure 2. Distributed Architecture (Reprinted with permission

from (The Computer-Based Patient Record: An Essential
Technology for Health Care, Revised Edition) # (1997) by the
National Academy of Sciences, courtesy of the National Academies
Press, Washington, D.C.)
also be difficulties related to each subsystem using

different data structures to manage and to store
records, making the design of an effective DBMS
potentially quite complex. These issues have been
successfully addressed in specific implementations of
distributed CPR systems.
DBMS and Media Processing
The general components of a CPR are shown in Fig. 3. An
ideal DBMS would be able to seamlessly integrate different
media formats (text, images, three-dimensional (3D) models, audio and video). Four important types of DBMS
commercially available are hierarchical, relational, textand object oriented. Each of these is better suited for one or
more different media and system architectures than the
others. Many modern CPR systems, especially those based
on a distributed architecture, use a combination of commercial DBMSs or utilize a custom DBMS.
Non-text media Electronic

images, pictorials, video and
audio data
(scanned documents, X-ray, CT,
MRI, audio notes)
External information data

generated outside the hospital (lab
measurements, referral letters
and ECG measurements)
CPR
Locally generated information
data generated in various
departments of the hospital
Unformatted text free text

(plans, findings and conclusions)
Formatted text organized text

(data in objective type fields)
Figure 3. Components of a CPR.
An important aspect of a DBMS is the mechanism that it

uses for data retrieval. The details of record construction
can play a significant role in creation of databases that can
be efficiently and accurately searched. Entries in formatted
fields can be searched much more easily than scanned
documents, images or nonformatted text. A DBMS should
also provide organization of the data in a manner that is
sensible and easy to comprehend. Well-organized databases will prove easier to manage and faster to search.
Security and Legal Issues
A core issue in the design of CPR systems is security. A
single security breach could be used to access thousands of
complete CPRs, which are then vulnerable to many malicious activities. Thus, it is extremely critical to protect
individual CPRs from unauthorized access, as well as to
protect the CPR system itself from hackers attempting to
access the system or attempting to cause it to fail. In
addition, the CPR data must be protected before it can
be sent across a network, due to potential communication
leaks that can land sensitive information in the wrong
hands. Tools such as network packet sniffers are extremely
powerful if an unauthorized party can achieve physical
access to network interconnection hardware or cabling. In
general, five conditions must be satisfied before patients or
health care personnel are granted access to the records on a
network (11):
1. The patient must not have recorded an explicit prohibition of CPR transmission across the network.
2. The patient must have consented to the transfer of
the partial or complete CPR to the recipient institution.
3. The identity of the patient must be accurately verified and authenticated.
4. The identities of the healthcare provider and
the recipient institution must be accurately
verified.
5. It must be verified that the patient is mentally
competent. If the patient is found to be mentally
incompetent, then the need for the patient record
must be sufficiently urgent that if it is not made
available immediately, serious harm could result to
the patients health.
For a CPR produced by a CPR system to be a legally
valid document, the CPR should be able to answer the
following questions (12).
1. Who Wrote It? A single CPR is frequently handled by
many different people, with several healthcare providers entering, updating, or removing information.
A CPR must include the identities of all the people
who have modified the record, together with the
information that they have modified. This involves
secure access control using smart cards, PINs, fingerprints, passwords, etc. to verify the identity of
requestors before they will be allowed to access or to
modify the information on the records.
357
2. When Was It Written? Time and date information is

maintained by automatically attaching timestamps
to the CPRs. The system must be designed to ensure
that unauthorized personnel cannot change the system time or any timestamp in any CPR.
3. What Does It Say? Computer technology is reinvented roughly every 5 years. Data storage and
retrieval media become obsolete very quickly. For
example, the magnetic tapes that were very popular
25 years ago can be found only in museums today.
It is thus important for CPR systems to incorporate
these technological developments, and yet not
become outmoded too quickly, so that the contents
of current CPRs and archived CPRs can both be read
and processed to obtain the same information.
4. Has the Information Been Altered? Undocumented
alteration of information will prove to be the most
important legal issue, and hence, the most
difficult one to solve. A CPR system must be designed
to allow determination of whether the information is
indeed as originally recorded or whether it has
been subsequently modified in a manner that
would affect possible legal proceedings. Several effective solutions, based on cryptographic techniques,
exist.
A very reliable solution to ensure that a particular CPR
provides answers to all of these four questions is to attach
digital signatures to them. Just as an ink-based signature
provides legal acceptability for paper-based records, digital
signatures can ensure the legal acceptability of a CPR.
Digital signatures offer both signatory and document
authentication, and they have been proven to be more
effective than the ink-based signature (13). Signer authentication provides the capability to identify the person who
signed the document, while document authentication
offers the capability to determine whether a document
was altered after it was signed.
Digital signatures use public-key encryption schemes to
provide this functionality. In public-key encryption
schemes (14), every user gets a pair of keys: a public
key, which everybody is allowed to know, and a private
key, which is kept as a secret known only by the individual
and the authority that issues and verifies the keys. A
digital signature is computed using the data bytes of the
document and both the private and public keys and
attached to the document. Consider that person A creates
a CPR and attaches a digital signature to it using his public
and private keys. Anybody with access to As public key can
verify that the CPR was actually created by A using the
CPR, the digital signature and As public key. If the result
is correct, according to a straightforward mathematical
relationship, A is authenticated as the signatory of the
CPR, and the CPR is authenticated as accurate. Any
alteration of the CPR after the calculation and attachment
of the digital signature would corrupt the digital signature
and would then cause the CPR to be identifiable as modified. In this case, the CPR would be considered fraudulent,
although the source of the fraud would not necessarily be
identifiable.
358
Public-key encryption can also be used to secure CPRs

before transmitting them across a network. Consider that
person A desires to send a CPR to person B. Person A looks
up the public key of B and encrypts the CPR using Bs
public key. The CPR can only be decrypted using Bs
private key. Since only B has access to his private key,
only he can decrypt the CPR. The principle of public-key
encryption can be extended to CPRs to protect the authenticity of the information, to identify malicious alterations of
the CPR, and to secure transmission across a network.
POINTERS TO THE FUTURE
Much of this article has been devoted to explaining the
benefits of the CPRs and the factors hindering their widespread implementation. Despite their tremendous potential, the development of commercial CPR systems has not
reflected the progress made in many related fields in
computer technology. In this section, some of these related
technologies that, in the future, the authors feel will have
great potential to broaden the range of advantages of
computer-based medical record services and to make them
sufficiently secure are presented.
Testing of replacement of bar codes with RFID tags in

patient bracelets is ongoing. Unlike bar codes, RFID tags
do not require clear line of sight between the bar code and
the bar code reader, nor do they require active operator
intervention. This ensures that healthcare workers will not
fail to scan a patient ID bracelet. One such successfully
tested system is Exaveras eShepherd, which combines
RFID tags with Wireless Fidelity (Wi-Fi) networks and
Voice over IP (VoIP) to implement a single system that will
track patients, staff and hospital assets (16). The Wi-Fi
routers can deliver patient information directly to a physicians handheld PDA every time any RFID transceiver
detects a patient. Physicians and hospital staff can have
the patient information whenever and wherever they
want, and they do not have to refer repeatedly to a secure
physical filing area to retrieve patient records.
A major factor impeding widespread implementation of
RFID tags is the legal and ethical issue of keeping such
detailed records of people, their activities, and all the
things they buy and use, without their consent. However,
once this issue has been resolved, RFID tags will change
the way patient records are processed.
Biometrics
Radio Frequency Identification Tags

Radio Frequency (RF) identification (RFID) refers to an
automatic ID system that uses small RF devices to identify
and to track individual people, pets, livestock, and commercial products. These systems are used to automatically
collect highway tolls and to control access to buildings,
offices and other nonpublic sites. An RFID tagging system
includes the tags themselves, a device to write information
to the tags, one or more devices to read the data from the
tags, and a computer system with appropriate software to
collect and to process information from the tag. Many
applications can use the same devices to both read and
write the RFID tags. While the physical layout of the tag
may vary according to the application, its basic components
will include an intelligent controller chip with some memory and an antenna to transmit and receive information.
According to its power requirements, an RFID tag will be
classified into one of two types. Active tags have their own
power source and hence provide greater range. Passive
tags will be powered by RF pulses from the tag reader (or
writer) and thus exhibit no shelf life issues due to battery
exhaustion.
While plans are underway to replace bar codes with
RFID tags on virtually every single commercial product,
the field of medical informatics has found some unique
applications for RFID tags. The U.S. Food and Drug
Administration (FDA) recently approved the implantation
of RFID tags on humans (15). These tags are similar to the
tags being used on animals, and they are implanted under
the skin in the triceps area. The chip contains a 16-digit
number that can be readily traced back to a database
containing the patients information. One such chip made
by VeriChip costs $125, exclusive of the cost of implantation. This technology is expected to be a boon to individuals
with life-threatening medical conditions and to lower medical costs by reducing errors in medical treatment.
Biometrics refers to automatic recognition of people based

on their distinctive anatomical and behavioral characteristics including facial structure, fingerprint, iris, retina,
DNA, hand geometry, signature, gait and even the chemical composition of sweat (17). Biometric systems have been
used in wide ranging applications like user authentication
before entry into buildings and offices, criminal investigation, and identification of human remains.
Biometric systems will also find an excellent application
in management of medical records. Individual traits of
people who are authorized to access the record can be
stored along with the record. When the system receives
a request from a user to retrieve the record, the system will
attempt to match the biometrics of the user to the database
of biometrics of all the people who are authorized to access
the record. The record is fetched only if the match is
positive. In this manner, a CPR can be viewed and modified
only by authorized personnel. This is potentially an effective application for biometric systems, because the system
can be trained with as many samples as needed from
authorized personnel. Biometric system applications
where a strong need exists to verify the authenticity of
claimed membership in a preidentified population have
been successfully tested with almost 100% accuracy. Such
systems also offer a deterrent to would-be intruders
because the biometric information associated with an
unsuccessful access attempt can be retained and used to
identify the intruder.
Generally, the biometric information of a user is
encoded on a smart card that the user must use to gain
access into the system. This provides a far better authentication solution than using just a user name and a password, because both of these can be forged fairly easily.
Most current biometric systems are based on the uniqueness of human fingerprints, and systems based on more
complicated biometric features are currently undergoing
investigation. The development of superior biometric systems holds the key to more secure CPR authentication.
Content-Based Image Retrieval
Content-Based Image Retrieval (CBIR) is the process of
retrieving images from a database based on the degree of
similarity in content between an example image (the query
image) and the various images contained in the database.
Traditional non-CBIR data retrieval mechanisms apply
text-based methods, in which a string query is matched
to user-generated descriptions of documents or to contents
of specific fields in a record. When patient images are
integrated into patient records, simple text-based searches
will not serve the same purpose. Text descriptions of
images are very subjective and require data entry by a
knowledgeable user. If that same image is to be added to a
different database, the accompanying description also
must be added if the image is to be retrievable.
These techniques utilize both global properties, such as
image statistics and color distributions, and local properties, such as position and texture, to retrieve images that
appear similar to the query image. This is a powerful tool
for physicians, because CBIR enables them to compare the
image of a current patient (query image) to similar images
of other patients in the database, to examine descriptions
attached to those images in their corresponding CPRs, and
to study the treatment administered in those cases, all in a
matter of seconds.
More and more frequently, scanned documents such as
referral letters and scanned paper-based records are being
added to CPRs. These documents can be treated as images,
and CBIR systems based on character recognition technologies can search for text descriptions on these documents,
making these documents compatible with text-based
searches. This technology can save a great deal of time,
money and errors in converting historical or ongoing
paper-based records to CPRs, because the records can
simply be scanned and need not be entered again into
the CPR system. As more and more visual aids are added
to medical diagnostics, CBIR will become indispensable to
CPR management.
359
patient details including other complications the patient

may have, and a referral letter, all of which are sensitive
and confidential information. The traditional method to
send the supplemental information electronically would be
to encrypt the information using public-key encryption
techniques and to transmit it as separate files together
with the image. If the transmission is intercepted by a
malicious party seeking private data, the transmission
would garner interest, because it would be quite obvious
that any encrypted information following the image information would likely be related to that image. The data thief
must still work to decrypt the transmission and to reveal
the confidential information. Decryption itself is difficult
without extensive computational capacity, but data thieves
will often find getting the necessary password to be a much
easier method to access the data. The disadvantage of
encryption is that the data thief knows what to try.
An alternate data transmission method would use steganography. The sensitive information can be embedded in
inconspicuous areas of the image (18). Thus, no additional
information or files are transmitted, and the image will
raise less suspicion that associated sensitive data can be
located. Although the embedded data is computationally
easier to decode than the encrypted messages, it will only
be decoded by people with foreknowledge that there is
information embedded in the image. Furthermore, the
embedded data can be encrypted before it is embedded,
so that the data thief has another level of barrier. The
information is much more difficult to locate, and the thief
still has to decrypt or to steal the password.
Usually, the clinically important sections of the image
will be identified, and only the remaining regions will be
used for embedding the secret information. Several sophisticated steganography techniques, like BPCS, ABCDE, and
modified ABCDE (1921) have been successfully implemented. These methods can hide significant amounts of
information without clinically altering the image, although
hiding too much data leaves evidence that can be detected
mathematically using tools from the field of steganalysis.
Figure 4 shows an example of data hiding using the
Steganography and Watermarking

Steganography and cryptography are related but distinct
methods that are used to ensure secure transmission and
secure archival of information. In cryptography, messages
are encrypted in such a manner that unauthorized recipients may not decrypt the messages easily, using long,
secure passwords and complex mathematical algorithms
to drastically alter the data. Often, however, the encrypted
messages themselves may be obtained fairly easily,
because they are transmitted over insecure networks
and archived on insecure servers. In steganography, the
very presence of secure information is masked by hiding
that information inside a much larger block of data, typically an image.
If an oncologist wishes to get a second opinion from a
gynecological specialist concerning a diagnosis of cervical
cancer, he might elect to send a digital image of the cervix
of the patient. He would also include his diagnosis, relevant
Figure 4. Steganography. (a) Original digital image of the cervix.

(b) Section of clinically unimportant segment of image in a. (c)
Section in b encoded with secret information.
360
modified ABCDE scheme. Figure 4a is a digital image of a

cervix, and Fig. 4b shows the bottom right section of this
image, which contains no clinically important information.
This section forms an 864432 RGB image, which requires
1,119,744 bytes to represent directly or 295,727 bytes to
represent when losslessly compressed to PNG format.
Figure 4c is the section in Fig. 4b after 216,326 bytes of
sensitive information have been embedded. This hidden
data corresponds to >40 pages of pure text, at 5000 characters per page. Figure 4c visually appears very similar to
the section in Fig. 4b, even with this extensive additional
data embedded. An even more secure solution would
encrypt the sensitive information before embedding it
inside the carrier image. This way, even if steganalysis
methods detected that the carrier image contains
embedded information, the decryption problem still
remains.
Watermarking techniques are similar to digital signatures in the sense that they can provide owner and document authentication for digital media. They are usually
nonvisibly detectable signals embedded into the media that
can be checked, by authorized personnel, to verify the
validity of the media and to trace its copyright holders.
Watermarking and steganography techniques can also be
used to identify times, dates, locations, and information
that could be used to cross-reference to a patient, so that
misidentification or misfiling of image data under the
wrong patient file can be detected and corrected. It is
the opinion of the authors that both these technologies
will provide valuable solutions for secure CPR systems.
CONCLUDING REMARKS
This article attempted to provide a brief overview of the
applications of computers in medical records, the benefits
of computerization, and issues concerning widespread
deployment of computer-based medical record systems.
The computerization of medical records is a vast area of
research, employing many people from diverse fields of
medicine, science and engineering. While we have
attempted to summarize and present the material from
our own perspective, the topic itself is explored in great
detail by several worthy publications. There are literally
thousands of publications that attempt to explain or to
solve one or more of the issues presented in this article and
to provide much greater detail than is possible in this
article. Thus, the authors felt it appropriate to leave the
reader with a few resources, both in print and on the World
Wide Web (WWW), to obtain more information about computers in medical records.
Much of the material in this article has been inspired
from the extensive discussions in Refs. 3 and 22. These
books provide interesting observations and refer to a
wealth of references that pioneered the computer revolution in the medical field. Two more recent books, (23 and
24), also provide excellent insight into the various aspects
of using computers to manage medical records. The WWW
contains many websites that provide solid insight into
various aspects of medical records management. Ref. 6 is
one such site that is constantly updated with information
about upgrading to CPR systems, cost benefit calculators,

software vendor surveys, and from basic tutorials on every
aspect of CPR management. Refs. 25 and 26 provide excellent evaluations of commercially available software for
CPR management.
ACKNOWLEDGMENT
The authors would like to thank Mr. Rodney Long at the
U.S. National Library of Medicine, Bethesda, Maryland,
for providing the images of the cervix used in the example
on steganography.
BIBLIOGRAPHY
1. Shortliffe EH. The evolution of electronic medical records.
Acad Med 1999;74(4):414419.
2. Burnum JF. The misinformation era: The fall of the medical
record. Ann Intern Med 1989;110:482484.
3. Dick RS, Steen EB, Detmer DE, editors. Institute of Medicine. The Computer-Based Patient RecordAn Essential
Technology for Health Care. Washington (DC): National
Academy Press; 1997.
4. Shortliffe EH, Perreault LE, Fagan LM, Wiederhold G. Medical Informatics Computer Applications in Health Care and
Biomedicine. 2nd ed. New York: Springer-Verlag; 2000.
5. Covell DG, Uman GC, Manning PR. Information needs in
office practice: Are they being met? Ann Intern Med 1985;
103:596599.
6. Voelker KG. (None). Electronic Medical Records [online].
http://www.emrupdate.com. Accessed 2005, April 10.
7. Weed LL. Medical records that guide and teach. New Engl J
Med 278(11):593600 and 278(12):652657.
8. Brailer DJ, Terasawa EL. Use and adoption of computerbased patient records, California Healthcare Foundation
report, Oct. 2003, ISBN 1-932064-54-0.
9. Wang SJ, et al. A cost-benefit analysis of electronic medical
records in primary care. Am J Med Apr 1 2003;114(5):397403.
10. Classen DC, Pestonik SL, Evans RS, Burke JP. Computerized surveillance of adverse drug events in hospital patients.
J Am Med Assoc 1991;266:28472851.
11. David MR, et al. Maintaining the confidentiality of medical
records shared over the Internet and the World Wide Web.
Ann Intern Med 1992;127(2):138141.
12. Cheong I. The legal acceptability of an electronic medical
record. Aust Fam Phys Jan. 1997;26(1).
13. Askew RA. Understanding Electronic Signatures [online]. Real
Legal. Available at http://www.reallegal.com/downloads/pdf/
ESigAskewWhitePaper.pdf. Accessed 2005. April 10.
14. Dent AW, Mitchell CJ. Users Guide to Cryptography and
Standards. Boston: Artech House; 2004.
15. Sullivan L. FDA approves RFID tags for humans. Inform
Week Oct. 2004.
16. Collins J. RFID remedy for medical errors. RFID J May 2004.
17. Jain AK, et al. Biometric: A grand challenge. IEEE Conference on Pattern Recognition, Vol. 2; Aug. 2004. pp 93526.
18. Johnson NF, Duric Z, Jajodia S. Information hiding: Steganography and WatermarkingAttacks and Countermeasures. Norwell (MA): Kluwer Academic; 2001.
19. Kawaguchi E, Eason RO. Principle and Applications of
BPCS-Steganography. Proc SPIE Int Symp Voice, Video Data
Communications; 1998.
20. Hirohisa H. A Data Embedding Method Using BPCS Principle
With New Complexity Measures. Proc Pacific Rim Workshop on
Digital Steganography. July 2002; p 3047.
MICROARRAYS
21. Srinivasan Y, et al. Secure Transmission of Medical Records
using High Capacity Steganography. Proc IEEE Conf
Comput.-Based Medical Systems; 2004. p 122127.
22. Dick RS, Steen EB, editors. Institute of Medicine, The Computer-Based Patient RecordAn Essential Technology for
Health Care. Washington (DC): National Academy Press; 1991.
23. Hartley CP, Jones III ED. EHR Implementation: A Step-by Step
Guide for the Medical Practice. Am Med Assoc Feb. 2005.
24. Carter JH. Electronic Medical Records: A Guide for Clinicians and Administrators. American College of Physicians
March 2001.
25. Rehm S, Kraft S. Electronic medical recordsThe FPM
vendor survey, Family Practice Management. Am Acad
Family Phys Jan. 2001.
26. Anderson MR. EMR frontrunners, Healthcare informatics
online, May 2003.
Thymine
Adenine
H
H2C
N H
INTRODUCTION
Microarrays allow for the simultaneous, parallel, interrogation of multiple biological analytes. Originally, microarrays were devised as a method by which gene expression
could be measured in a massively parallel manner (all the
genes in the genome at once), however, recent advances
have demonstrated that microarrays can be used to interrogate epigenetic phenomena, promoter binding, protein
expression, and protein binding among other processes.
The overall process is reliant upon the manufacture of a
highly ordered array of biological molecules, which are
typically known entities. The features of this array behave
as probes, which react with and bind to the unknown, but
complimentary material present in a biological sample.
Here we will focus specifically on gene expression (deoxyribonucleic acid, DNA) microarrays, which can be used to
assay the activity of thousands of genes at a time.
In 1993, Affymetrix published a novel method of using
light directed synthesis to build oligonucletide arrays that
could be used for a variety of biological applications (1).
Shortly thereafter, a group lead by Patrick Brown and Ron
Davis at Stanford University demonstrated that robotically printed cDNA arrays could be used to assay gene
expression (2). Now, more than a decade after this initial
work was made public, both types of DNA array are commonly found in genomics laboratories.
BASIC PRINCIPLES
A DNA microarray contains a highly ordered arrangement
(array) of several discrete probe molecules. Generally, the
N
N
N
N
To DNA Chain
Cytosine
Guanine
H
N H
N
N
H N
N
O
To DNA Chain
NEIL WINEGARDEN
University Health Network
Microarray Centre, Toronto
Ontario, Canada
H N
See also EQUIPMENT ACQUISITION; OFFICE AUTOMATION SYSTEMS; PICTURE

ARCHIVING AND COMMUNICATION SYSTEMS; RADIOLOGY INFORMATION
SYSTEMS.
MICROARRAYS
361
H N
H
Figure 1. WatsonCrick base pairing interactions. During

hybridization, specific base-paring interactions occur by which
Thymine (T) binds specificly to Adenine (A) and Cytosine (C)
binds specifically to Guanine (G). The binding of these bases to
one another is mediated by hydrogen bonding as shown. The GC
base pairs are stronger by virtue of the three hydrogen bonds
formed compard to only two for AT.
identity of these probes, be they cDNA or oligonucleotides, is

either known or can be determined readily. The probes are
deposited by some means (see the section Fabrication of
Microarrays) onto a solid-support substrate such as glass or
silicon. DNA microarrays take advantage of a basic characteristic of DNA, namely, the ability of one strand of DNA
to find its complementary strand in solution and bind
(hybridize) to it. This hybridization event is highly specific
following standard WatsonCrick base pairing rules (Fig. 1).
Gene Expression
With some exceptions, the genetic makeup of every cell in
an organism is the same. Each cell has the same complement of genes, which comprise the organisms genome. The
subset of genes that are active in a particular cell dictate
that cells function. When we say a gene is active or
expressed, we mean that particular gene is being transcribed. Transcription is the process by which ribonucleic
acid (RNA) polymerase II (an enzymatic complex) reads a
gene and creates a complementary copy of messenger RNA
(mRNA). The more a gene is transcribed, the more copies of
mRNA will be present in a cell. Thus genes that are highly
active in the cell will be represented by multiple copies of
mRNA, whereas genes that are inactive in the cell will have
very few or no copies of mRNA in the cell. Microarrays
function to measure the amount of mRNA present in the
cells of a biological sample such as a tumor biopsy. The
activity of the genes is inferred from this measure.
Gene Structure
In higher eukaryotes, somatic cells (diploid) have
two copies of every gene: one maternally and the other
362
MICROARRAYS
paternally derived. In the context of the diploid cell, each

copy is termed an allele. In the case where both inherited
alleles are the same for a given gene, that gene is said to be
homozygous. If the two alleles are different, then the gene
is heterozygous. Alleles may be either dominant (phenotypically manifested regardless of what the other allele is),
or recessive (phenotypically manifested only in the absence
of a dominant allele). In the case of a heterozygous gene,
the dominant allele will be phenotypically manifested and
the recessive allele will not. If both alleles are different, but
dominant, they are termed codominant and both alleles
will elicit a phenotype. The gene is comprised of DNA,
which is double stranded. One strand is the sense strand or
the strand that encodes the information, which will be
ultimately represented in mRNA. The other strand is said
to be anti-sense and is the strand of DNA that is actually
read by the RNA polymerase to generate the mRNA. DNA
has directionality: A gene is transcribed starting at the 30
end of the antisense strand of the DNA and is read toward
the 50 end. The resultant mRNA is made from the 50 to the 30
end.
Genes are regulated by specific sequences of DNA that
lie outside the coding region of the gene. The first such
sequence is the promoter. Promoters bind the transcriptional machinery (RNA polymerase II) that performs transcription. Promoters are found 50 (upstream) of the gene
and are proximal to the transcription start site. An additional class of regulatory sequence called an enhancer may
be associated with the gene. Enhancers may lie upstream,
downstream, or internal (usually in noncoding regions
termed introns) to the gene (3). Specific transcription
factors bind enhancers and promote recruitment or activation of the basal transcriptional machinery. It is the coordinated function of the promoter and enhancer, with the
transcription factors that bind them, that control if a gene
is active or not within the cell. Thus, genes are regulated,
and can be turned on, off, or modulated up or down by the
regulatory mechanisms of the cell.
RNA Isolation
Ribonucleic acid must be isolated from cells in order to
prepare the material for hybridization to the array. A cell
contains three major species of RNA: mRNA, transfer RNA
(tRNA), and ribosomal RNA (rRNA). Together they are
refered to as total RNA. For the purpose of gene expression
experiments with microarrays, the mRNA is the species we
are interested in and represents 1% of total RNA. In
order to isolate total RNA from cells, one of two main
modalities is used: solution- or solid-phase extraction. In
solution-phase methods, cells are lysed in the presence of
isothiocyanate in order to inactivate any RNases (naturally
occurring enzymes that nonspecifically degrade RNA). The
lysate is then extracted with an acidified phenol:
chlorophorm:isoamyl alcohol solution. The RNA selectively
partitions to the aqueous phase of this mixture away from
proteins and DNA. The aqueous phase is removed and RNA
is precipitated out of solution using isopropyl alcohol at
high salt concentrations. Solid-phase methods make use of
the variable binding activity of RNA to a silica matrix at
high and low salt conditions. Cells are again lysed in the
presence of isothiocyanate. The high concentration of isothiocyante used in this methodology not only inactivates
the RNases, it also selectively precipitates proteins out of
solution. The lysate is applied to a column containing a
silica filter at the bottom. The lysate is pulled through the
column via vacuum or centrifugation, thereby removing
the proteins and cellular debris. In this method, DNA may
also bind to the column, and as such contaminating DNA is
removed by the application of DNase. The column is
washed to remove any further contaminants, and then
the RNA is eluted from the filter using water.
mRNA Structure
In eukaryotic cells, mRNA has a unique feature that allows
researchers to either purify it away from the rest of the
RNA or to direct enzymes to it specifically while avoiding
the other RNA species. This feature is the polyA tail. The
polyA tail is a long stretch of adenine nucleotides found at
the 30 end of mRNA, which is added post-transcriptionally.
Such stretches of adenine nucleotides do not typically occur
naturally in genes or other RNA species. The polyA tail will
hybridize to an artificially generated oligonucleotide made
up of a series of deoxythymine nucleotides (oligo-dT). If the
oligo-dT is coupled to a support matrix (e.g., beads) the
mRNA can be pulled out of solution thereby purifying it
away from the rest of the total RNA. While some researchers prefer to include this step in their process, it is generally not a requirement for microarray analysis. Rather
than purify the mRNA, the oligo-dT can be used as a primer
for creating an enzymatically labeled complement of the
mRNA.
Labeling
In order to render the RNA visible to a detection system, it
is necessary to label it in some manner. While some
laboratories choose a direct methodology of chemically
labeling the mRNA itself, it is most common to work via
a cDNA or cRNA intermediate that is labeled enzymatically.
The simplest methodology involves creating labeled
cDNA. In this technique, the RNA is reverse-transcribed
(DNA is made from an RNA template) by an enzyme named
reverse transcriptase (RT) (for sample protocols, see Ref.
4). Reverse transcriptase requires a small oligonuclotide
primer that binds to the RNA creating a short doublestranded region (an RNA:DNA hybrid). In order to ensure
that the RT enzyme reads only the mRNA, the polyA tail of
mRNA is exploited by using a primer made of a stretch of
several (usually 2025) thymine residues. The resultant
DNA is the complement of the RNA and it is thus referred
to as complementary DNA (cDNA). The RT reaction
requires that free nucleotides (each of A, C, G, and T)
are present to create the DNA. If one of these nucleotides
is chemically modified with some detectable molecule (such
as a fluorophore), then it will be incorporated into the
cDNA strand, and that cDNA will be detectable with a
fluorescent reader. Alternatively, it is possible to use a
reactive molecule (such as amino-allyl) in place of a fluorescent molecule. After incorporation into the DNA, the
DNA is then coupled to a reactive form of a fluorophore
MICROARRAYS
(usually a reactive ester). This latter implementation of the

method has an advantage in that the amino-allyl modifier
is a much smaller chemical group that is incorporated
much more efficiently into DNA than a bulky fluorescent
moiety.
Often the amount of RNA available is limiting and
cannot be detected by standard means. In this case, it is
generally necessary to amplify the amount of material
present. A typical microarray experiment usually requires
510 mg of total RNA in order to be able to obtain useful
data. When researchers are working with diminishingly
small samples, such as from a needle biopsy or a fine needle
aspirate, it is often not possible to obtain this amount of
total RNA. To overcome this limitation, various amplification strategies have been adopted. The most popular
method of amplification is based on the protocols of Dr.
James Eberwine from the University of Pennsylvania (5).
In this technique, RNA is converted into cDNA using the
same method described above with two key differences: (1)
there is no labeled nucleotide incorporated and (2) the
oligo-dT primer has another short sequence of DNA
appended to it that represents a T7 promoter region.
The T7 promoter is a bacteriophage-derived sequence that
initiates transcription by T7 polymerase. After the cDNA is
created, a second strand is generated creating a doublestranded artificial gene with a T7 promoter on one end.
This artificial gene is then transcribed by the addition of T7
polymerase, which is allowed to make numerous transcripts of the gene. The transcripts that are obtained can
either be labeled directly, or they in turn can be turned into
labeled cDNA using standard methodologies described
above. The resultant RNA is now actually the opposite
sequence of the original mRNA, so it is said to be cRNA
(complementary RNA).
The Affymetrix GeneChips utilize an amplification system based on T7 transcription as described above. During
the production of cRNA, biotin modified nucleotides are
incorporated. Posthybridization (see the section on Hybridization) the arrays are stained with a streptavidin bound
fluorophore. Streptavidin is a protein that specifically and
tightly binds to biotin molecules, allowing the fluorophore
to be attached to the cRNA.
A clean-up step is required to remove any free, unbound
detection molecules. This step helps to ensure that background signal is kept to a minimum. There are two main
methods by which such purification is performed, one is
based on standard nucleic acid purification systems, similar to the RNA isolation method described earlier, and the
other is based on size exclusion. For the first method, a
nucleic acid purification column is utilized. The cRNA or
cDNA binds to the silica filter, but the less charged free
nucleotides flow through. After a series of washes, the
cRNA or cDNA is eluted from the column. The second
methodology utilizes a membrane filter (usually incorporated into a column) that has a defined pore size. The large
cRNA and cDNA molecules are retained on the membrane;
where as the small free nucleotides flow through. The
column is then inverted and the cDNA or cRNA is then
eluted off the column by flowing wash buffer in the opposite
direction. This purified labeled material is then ready for
hybridization to the array.
363
Hybridization
Microarray technology relies on the natural ability of
single-stranded nucleic acids to find and specifically bind
complementary sequences. Purified labeled material is
exposed to the spotted microarray and the pool of labeled
material self-assembles onto the array, with each individual nucleic acid (cDNA or cRNA) species hybridizing to a
specific spot on the array containing its complement. The
specificity of this interaction needs to be controlled, as
there may be several similar and related sequences present
on the array. The control of hybridization specificity is
accomplished through the adjustment of the hybridization
stringency. Highly stringent conditions promote exact
matches where as low stringency will allow some related,
but nonexact matches to occur. In a microarray experiment, stringency is typically controlled by two factors: the
concentration of salt in the hybridization solution and the
temperature at which hybridization is allowed to occur.
High salt concentrations tend to lead to lower stringency of hybridization. Both strands of nucleic acid
involved in the hybridization event contain a net negative
charge. As such, there is a small repulsion between these
two strands, which needs to be overcome to bring the
labeled nucleic acid into proximity of the arrayed probe.
The salt ions cluster around the nucleic acid strands creating a mask and shielding the electrostatic forces. Higher
salt concentrations have a greater masking effect, thus
allowing hybridization to occur more easily. If salt concentrations are high enough, the repulsion effects are completely masked and even strands of DNA that have low
degrees of homology may bind to one another.
Temperature is another important factor. Every doublestranded nucleotide has a specific temperature at which
the two strands will melt or separate. The temperature at
which exactly 50% of a population of pure double-stranded
material separates is termed the melting temperature
(Tm). The Tm of a nucleic acid is controlled partially by
the length of the strand and partially by the percentage of
G and C residues (termed the GC content). The G and C
residues bind to one another as a WatsonCrick base pair.
This pairing interaction is the result of three hydrogen
bonds forming. The other potential base pair in a DNA
hybrid, A:T, only has two such hydrogen bonds and thus
the greater the GC content of the nucleotide, the more
stable the hybrid. At very low temperatures, nonstandard
WatsonCrick base pair interactions can also occur causing
noncomplementary sequences or sequences that are
<100% matched to form hybrids. It is necessary therefore
to find a temperature that will prevent or melt nonspecific
hybrids, but allow the specific interactions to occur. For a
microarray, this presents a challenge as there are thousands of specific interactions that must be accommodated.
In the case of oligonucleotide arrays, the design of the
oligonucleotides to be spotted takes this issue into account
and probes are designed that tend to fall within a narrow
window of potential melting temperatures. cDNA arrays
are more difficult because the sequences spotted vary
greatly in both GC content and length. In such cases, it
is often true that conditions that represent somewhat of a
compromise are necessary.
364
MICROARRAYS
Hybridization kinetics can generally be modeled as

shown in Eq. 1(6). The change in the amount of hybridization product LS over time is a function of the decrease in
the concentration of labeled target L and free spotted DNA
S over time. To simplify the equation, the rate of hybridization is equal to some rate constant k multiplied by the
product of the concentrations of L and S. Thus hybridization rate is a direct function of the concentrations of the
labeled target molecule and the DNA probe in the spot.
dLS
dL dS dL S

k LS
dT
dT
dT
dT
In the case of an oligonucleotide microarray, it is often the

case that the number of spotted DNA molecules is in great
excess to the number of target molecules. As such, the
concentration of the spotted DNA probe remains fairly
constant and can be considered part of the constant k.
Thus the equation for hybridization can be simplified as
shown in Eq. 2 (6), where the rate of hybridization is
typically driven by the concentration of the labeled target
molecules alone.
dLS
2
k0 L
dT
In the case of two color oligonucleotide arrays, the two
labeled samples compete for hybridization to the probe that
remains in excess and thus hybridization is simply a
reflection of the concentrations of each of the two labeled
targets L1 and L2 [Eq. 3(6)].
dL1 S k01 L1
dL2 S k02 L2
The situation becomes somewhat more complex when the

probe molecules are not in excess of the target molecules.
This is often the case with cDNA arrays. In these cases, the
concentration of the spotted probe does change significantly as hybridization occurs and thus each of the labeled
targets L1 and L2 hybridize in a manner described by Eqs. 4
and 5 (7).
dL1 S
k1 SL1 k1 S0 L1 S L2 SL01 L1 S
dT
4
dL2 S
k2 SL2 k2 S0 L2 S L1 SL02 L2 S
dT
5
In such a case, the rate of hybridization is affected by the
change in the concentrations of the spotted probe from the
initial concentration S0, where S0 changes as the probe
molecules are bound by either L1 and L2.
When looking at differential hybridization between the
two targets, we can represent the kinetics as shown in
Eq. 6 (7).
dL1 S k1 L01 L1 S
6
dL2 S k2 L02 L2 S
If one is to assume that the two fluorescent molecules
used in a two-color experiment behave similarly, and that
the rate of hybridization of the two labeled targets is the
same, we can say k1 k2. It has been demonstrated that
under ideal conditions and when the hybridization reaction
is allowed to continue to equilibrium that the ratio of the

concentrations of each possible hybrid L1S and L2S is
equivalent to the ratio of the original concentrations of
the two targets L1 and L2 [Eq. 7 (7)]. This point is important
because it is the basis for microarrays to work, assuming
that the ratios read from the scans during data analysis are
reflective of an actual biological condition.
L1 S L01
L2 S L02
The goal of microarray hybridization is to produce a

result for which the signal obtained from specific hybridization is very strong when compared to any background
signal that may be obtained by a nonspecific adsorption of
labeled material to the substrate, or nonspecific binding to
spotted elements. To reach this goal, it is common to use
certain nonspecific blocking reagents in the hybridization
solution. Frequently, nucleic acids from sources known not
to contain any sequences that will interfere with specific
hybridization are used. For example, in a hybridization of a
human sample to an array, one might use yeast tRNA and
salmon sperm RNA as competitors to bind any regions of
the substrate or probes that have a generic nucleic acid
binding capacity. These nucleic acids are nonlabeled and
will therefore not contribute any signal when the array is
scanned.
Washing
Unlike traditional northern blots, the majority of the
stringency of a microarray assay is accomplished at the
hybridization step. The washing step of a microarray
experiment is a critical operation, but is important more
as a means to remove unbound material in order to reduce
background signal than it is to control the specificity of the
signal obtained.
Wash buffers generally contain two components: a salt
solution and a detergent. The salt solution, frequently
sodium chloride sodium citrate (SSC), is set to a concentration that supports the maintenance of the hybridized
molecules. This concentration most frequently falls in the
1 to 2 concentration range with some labs using as low
of a concentration as 0.1 (1 SSC contains 0.15 M NaCl
and 0.015 M Na-citrate).
The detergents used in wash buffers help to remove
the unbound fluorescent molecules that would normally
stick to the surface of the slide. The detergent acts as a
surfactant and helps to isolate and remove the unbound
fluorescent material. Typically, an anionic detergent
such as sodium dodecyl sulfate (SDS) is used for this
purpose.
The temperature for the washes varies depending on the
stringency of the wash solution being used. As with hybridization, the combination of temperature and salt concentration determines the overall stringency of the washes.
After washing the microarrays, it is generally necessary
to perform a rinse. The rinse is typically a solution similar
to the wash solutions without the detergent. If detergent
remains on the slide after drying, the solution may fluoresce particularly if the labeled material has been trapped in
detergent micelles.
MICROARRAYS
Scanning
It is necessary to use an imaging device to detect the
fluorescent labels present on the hybridized microarray.
In general, the imaging device must contain an excitation
light source, an emission filter, and a light gathering
device.
During scanning, the labeled material, be it fluorescent
or some other form of detectable molecule, is imaged and
the resultant data is converted to a digital image. The
optimal resolution at which the image is scanned is dependent on the size of the features and on their interspot
spacing. A general rule of thumb is that the resolution of
the image should be such that the pixels represent onetenth of the diameter of the spot. For spotted arrays, for
example, the features tend to be on the order of 100 mm in
diameter and thus 10 mm resolution is frequently used.
Affymetrixs technology, however, can generate features
that are 11 mm square; in this case, a much higher resolution of down to 1 mm is required.
Most commonly, the image that is generated is a 16-bit
grayscale TIFF (Tagged Image File Format) image
(Fig. 2). The 16-bit depth of the image provides a total
of 65,536 gray levels providing a possibility of more than
five orders of magnitude range. The TIFF format is important because it is a universally accepted format that is
LOSSLESS; that is, even with compression, this format
retains all image information. The images can then
be imported into the appropriate image quantification
software.
Image Quantification
After scanning, it is necessary to extract data from the
images. Image quantification generally starts with
segmentation. Segmentation is the process by which
pixels that represent the signal are isolated from those
that represent background. During segmentation, the
discrete areas of the image that represent the spotted
DNA material are identified and digitally isolated from
the remainder of the image. The intensities of all of the
365
pixels in the individual spot are averaged to determine

the overall spot intensity. This spot intensity is proportional to the amount of material hybridized to that
region, with higher intensities resulting from increased
numbers of hybridized molecules. Each spot, for each
channel (in the case of two color microarrays) is
quantified, and the resultant data are tabulated. Other
data may also be extracted at this stage. It is common to
also obtain intensity data for the area outside of the
individual spots. This value represents the background
of the image and indicates the amount of signal that would
have been obtained regardless of a specific hybridization
event. It is common, however, not universal, to subtract
the background values from the signal intensities of the
spots.
There are several means by which segmentation can be
carried out. In the most basic setup, a fixed shape (usually a
circle) is placed over each spot. The entire complement of
pixels lying within the circle is used to determine the
average intensity. Pixels lying outside of one of these
circles are deemed to be background signal. More advanced
segmentation algorithms attempt to account for the fact
that most of the spotted features on a microarray are in fact
not perfectly uniform. Spots may deviate from a true
circular shape, or may have regions within the circle in
which DNA was not attached (creating a spot that is
reminiscent of a doughnut). In addition, it is not uncommon
for each of the spots to have some degree of variance in
their diameter. The more advanced methods utilize various
algorithms and statistics to determine which pixels actually represent signal and which are more representative of
background.
Image quantitation software then processes the entire
image and produces a table of results that represents the
signal, and the background for each feature on the array.
These packages may also export various other data, which
can be used in quality control analysis such as standard
deviations, coefficients of variance, circularity, or uniformity of the spot, and so on. This data table can then be
processed as part of the data analysis.
Figure 2. Arrays imaged on a

microarray scanner are presented as
16-bit grayscale TIFF images. The
picture shown represents a small
subsection of a larger array. Each
spot is 100 mm in diameter and the
spot-to-spot spacing is 200 mm in this
image. The image was scanned at
10-mm resolution.
366
MICROARRAYS
Data Analysis
An exhaustive description of the process of DNA microarray data analysis is far beyond the scope of this article
(for an excellent review see Ref. 8). The exact process
followed depends greatly on the experimental design and
the question being addressed. There are, however, some
basic principles that tend to be fairly common in dealing
with microarray data: statistical analysis of data, supervised and/or nonsupervised data mining, data visualization, and validation are all key components.
Statistical analysis of microarray data comes into play
in two main areas. The first is to determine which spots are
reliable and provide sufficient data. Spots that have a high
degree of variance across replicates, for example, are likely
not able to provide reliable data. These hypervariable
genes or signals need to be filtered from the data so as
to not skew the results of data mining. Statistics may also
play a role in supervised data analysis.
There are two major categories of data mining: supervised and nonsupervised. Supervised data mining utilizes
algorithms in which the user imparts restrictions on how
the data is grouped. For example, in an experiment where a
cohort of patients was tested in which one group
was healthy and the other group was afflicted with a
particular disease, one would indicate to the algorithm
which arrays were from the healthy patients and which
were from the patients with disease. The algorithm then
tests the data to find genes that are markers for the
diseases. Specifically, each gene is tested to see if the
expression levels for that gene are statistically significantly different in each of the two patient groups. The goal
is to find a series of genes that can act as markers that are
diagnostic of the disease.
In nonsupervised clustering, the algorithm is not given
any indication as to how the individual samples are
related. In true nonsupervised clustering, the algorithm
is not even told how many groups exist. The data are
analyzed and the samples are grouped based on similarity
metrics. The classical methods of nonsupervised clustering
include hierarchical clustering and principal components
analysis (PCA). The algorithms generally display the data
via some visualization pattern such as the canonical
plaid expression patterns seen from hierarchical clustering. The researcher then overlays the grouping information onto the patterns provided to see if the individual
groups naturally separate from one another. In other cases,
this methodology may be being used to determine how
many groups there truly are, as the researcher may not
have this information a priori. In such cases, the groups
can then be further examined to see if there are differences
in treatment response, survival, or any other characteristic
desired. Generally, after this technique is performed one
will attempt to look for clusters of genes in the patterns
that distinguish between the different groups and again
use these genes as markers.
Regardless of the methodology utilized, it is extremely
important to validate the data. Cross-validation strategies
are various, but in their most basic form, one obtains a
cohort of patients to profile. A subset of this cohort is used
to look for potential markers. Once the markers have been
identified, the remaining patients are tested and only the

identified markers are used to try and group the patients. If
the markers are able to stratify the patients into their
appropriate groups, then the markers are considered to be
viable and may provide beneficial diagnostic ability. On
occasion, however, the validation set is not properly
grouped. In such cases, the markers are only useful for
the narrow set of patients used in the initial tests and more
testing is required to find a viable set of markers.
FABRICATION OF MICROARRAYS
There are two main methodologies for manufacturing
microarrays, which differ in the means by which the probe
material spotted onto the arrays is prepared. In one methodology, the DNA to be spotted is generated in situ using
either standard or modified phosphoramidite chemistry.
(Phospohoramidites are reactive forms of each of the
nucleotides that make up DNA. Phosphoramidite chemistry is a well-defined process by which moderate length
stretches of DNA can be created with any specific
sequence.) This method is used by Affymetrix and Agilent,
the two largest commercial suppliers of microarrays,
although both groups use a different approach to the
in situ synthesis.
Other groups use ex situ synthesis, whereby the DNA
material is either prepared as PCR products (cDNA) or
oligonucleotides manufactured using standard phosphoramidite synthesis. Once this material is prepared it is
spotted onto the array substrate using either contact or
noncontact printing methodologies. Amersham (now GE
Healthcare) and Applied Biosystems use this methodology
to make microarrays as do almost all of the homebrew
laboratories that make microarrays in house.
Fabrication of DNA Arrays In Situ
There are two main approaches to the generation of microarrays by in situ synthesis of DNA: photolithography and
inkjetting. Affymetrix, the industry leader uses a proprietary photolithography process to mask off areas of the
array, protecting some areas, and leaving others available
for the DNA synthesis reaction to occur (1). This is a
multistep process requiring several masks per array to
be made. Each synthesis reaction is performed sequentially. For each nucleotide position, there are four possible
masks (one for each of A, G, C, and T). Thus, an array
comprised of 25-mer oligonucleotides would require 100
masks to complete the process (typically 70 are required
for an array due to the sequences used). Affymetrix uses a
modified phosphoramidite chemistry for synthesis of the
oligonucleotide chains; whereas standard phospohoramidite chemistry uses acid labile protection groups, the Affymetrix technology utilizes groups that can be removed by
ultraviolet (UV) light. The Affymetrix technology allows for
extremely high density arrays of hundreds of thousands of
features to be prepared on very small substrates of <1 cm2.
Other groups have developed technologies that allow
them to get around the need for multiple masks to be
made for each array design. The pioneer in this area
was Nimblegen, who uses digital light processor (DLP)
MICROARRAYS
micromirrors to create the masks (9). Each of these DLP

units (used typically in AV projectors and large screen
televisions) comprises thousands of tiny (10 mm2) micromirrors. The micromirrors can be individually addressed
and the angle of the mirrors changed to allow light to pass
through. In the open state, the micromirror directs light
onto the surface of the microarray, allowing DNA synthesis
to occur. In the closed state, the micromirror reflects light
away from the surface, disallowing DNA synthesis. A
computer controls the mirrors and thus each DLP unit
has a near infinite number of combinations that can each
be controlled, and as such, a single unit can create any
pattern desired on the array. Nimblegen uses the same
chemistry as Affymetrix, using light activated deprotection
of the phosphoramidites. A somewhat newer entry into this
area is Xeotron (now part of Invitrogen). Xeotron also uses
micromirror DLPs to address the masks, however, they
have also incorporated small microfluidic channels on their
chips. Each feature is placed in a microscopic well on the
chip. Rather than using the modified phosphoramidite
chemistry of Affymetrix and Nimblegen, Xeotron uses
standard chemistry, but has instead employed a caged acid
that can be freed by light (10,11). As such, the acid that
controls deprotection of the nascent oligonucleotide can be
directed to specific locations by light. The Nimblegen and
Xeotron technologies have the advantage of being highly
amenable to custom array generation, however the Affymetrix technology is particularly well suited to mass production of a standard array. Each of these approaches has
found customers in the marketplace.
A third approach to in situ synthesis of the oligonucleotides involves ink-jet spotting. Agilent uses this technology
(developed by Rosetta Inpharmatics) in which each of the
reactive phosphoramidites (A, G, C, and T) are loaded in to
a separate ink-cartridge to allow for control of which
nucleotide is added to each spot during the synthesis stage
(12,13). This methodology eliminates the need for masks,
but does require very high precision robotics as the print
head must return to the same spot many times, within
micron accuracy, during the course of synthesis. This
technology draws from the strength of each of the others
mentioned in that it is relatively easy to customize the
design of arrays, and yet, mass production of arrays is
possible using a large robotic system.
Fabrication of DNA Arrays Ex Situ
Some of the commercial vendors and nearly all of the
homebrew microarray centers utilize and approach of
spotting DNA that was prepared ex situ. In the case of
cDNA arrays, the spotted material is prepared by polymerase chain reaction (PCR), whereas oligonucleotide
arrays are generated using oligos created via high throughput oligo synthesis. The DNA material is purified and
placed into a specific spotting buffer that is compatible
with the substrates being used.
The DNA is typically aliquoted out into multiwell plates
(96, 384, or 1536 wells /plate) to facilitate transfer by the
arraying robot. The buffer that the DNA is placed in has
several functions. First, the buffer stabilizes the DNA to
prevent it from degradation. Second, the buffer must
367
provide an appropriate surface tension to ensure that

the spots that are placed on the substrate are of a controllable size and uniform in shape. Of similar importance,
however, is that the buffer must provide conditions that are
compatible with the attachment chemistry that is going to
be utilized.
The DNA may either be coupled to the slide through
rather simple electrostatic interactions or via a specific
coupling reaction. Electrostatic interactions are mediated
by using a uniform positively charged substrate that
attracts the negatively charged DNA. Often the substrates
used are silylated to provide reactive amine groups on the
surface. Alternatively, one may coat the slides with a
chemical such as poly-L-lysine, which simply adsorbs onto
the substrate and provides a net positive charge. This type
of interaction is mass based. As such, there is a maximum
mass of DNA that can bind to any one spot on the substrate.
Longer DNAs will be represented by fewer copies than
shorter DNAs. To overcome this, it is possible to use more
specific interactions by using modifiers on the DNA that
will react with certain groups on the slide. The two most
common such modalities involve aldehyde or epoxide chemistry. In this method, the DNA is modified with a primary
amine group. The substrate has reactive aldehydes or
epoxides that will react specifically with the primary amine
to form a covalent bond (Fig. 3). This type of interaction is
molarity based, and as such, with the exception of steric
effects, the number of DNAs that bind per spot is relatively
equivalent regardless of length.
EQUIPMENT
The manufacture of microarrays, and their subsequent use
requires some very specialized equipment. Generally, a
facility that produces microarrays will require some
advanced robotics for fabrication. A laboratory that uses
arrays will require scanning devices to read the arrays.
Due to the relatively high costs of these pieces of equipment
it is common for many people to rely on core facilities for
some or all of the process.
Arraying Robots
Ex situ prepared DNAs are spotted onto the microarray
substrates via robotics (Fig. 4). Robotics are required to
accurately position the printing devices over the slides to
create the arrays. The majority of systems utilize pins and
direct contact to deposit the DNA material. In this system,
a printhead with several spotting pins in a defined arrangement is used to dip into the multiwell plates and pick up the
material to be spotted. The typical operation sequence of an
arrayer robot may include:
1. Dipping the printing applicators (pins) into a source
plate to pick up DNA samples. Each applicator picks
up a separate DNA sample from an individual well in
the plate. Typically 3248 pins are used at one time.
2. Movement to a blot-station to preprint from the pins.
This step removes excess solution from the pins to
ensure that the spots that are printed onto the arrays
368
MICROARRAYS
(a)
(b)
NaBH4
C
Slide
H
NH DNA
C
Slide
NH2 DNA
H
H
C
N+H2-DNA
Slide
H
(c)
O
HC
Proton Transfer
CH2
Slide
NH2
DNA
OH
C
Slide
H
NH-DNA
Carbinolamine
O
HC
Slide
CH2
N+H2
DNA
H2O
O+H2
C
Slide
H
NH-DNA
HO
HC
Slide
CH2
NH
DNA
H2O
DNA
N+
OH
C
Slide
Iminium Ion
Figure 3. Covalent attachment of aminomodified DNAs to aledhyde (a) or epoxide (b)

slides is possible. An amino-modified DNA
reacts with an aldehyde surface by a Schiffs
base reaction. The resultant Schiff base must
be reduced with an agent such as sodium
borohydride (NaBH4) to prevent reversal of the
reaction.
DNA
N
+ H3O
C
Slide
Imine (Schiff Base)
are uniform in size and do not run into one another

causing contamination.
3. Movement to the slide platform. The print head then
moves over the slide platform taking position over
the first slide.
4. Printing onto the arrays. The print head moves down
bringing the pins in contact with the slide. The DNA
solution held in the pins by capillary action is spotted
onto the slide. The printhead then moves to the next
slide position and again spots onto the slide. This
process is repeated until all of the slides on the

platform have been printed.
5. Washing the pins. The print head then moves the
pins to a wash station. Although there are many
configurations possible, the basic principle is to use
water or some other solution to remove the excess
liquid from the pins and then to dry the pins (under
vacuum or stream of air). This process may be
repeated several times to make sure there is no
carryover.
MICROARRAYS
369
rotation creating a mixing effect. The fluidics station is a

more advanced system that is required to introduce the
various labeling components and wash solutions required.
This station allows the user to keep the cartridge sealed
without having to attempt to pipette solutions in and out.
Scanners
Figure 4. A microarraying robot. The robotic arrayer prints DNA

onto glass slides with very high precision. Robots such as this have
extremely high accuracy, on the order of 10 mm or less.
6. Loading the next sample. The print head returns to

the source plate to pick up the next set of samples.
In a typical high throughput system, such as those

offered by Bio-Rad, BioRobotics, GeneMachines, Genetix,
and Telechem International, 48 pins are used at one time.
The entire operation sequence described above may take
34 min to complete for 100 arrays. Often arrays may
contain 20,00040,000 spots. As such, a typical print run
may require 600 or more cycles through the operation
sequence, which can take as long as 30 h or more to
complete.
Hybridization and Fluidics Stations
Certain array platforms require that a specific hybridization and/or fluidics station be utilized. In the case of spotted
arrays (home-brew in particular), this is usually an option
and often a case of personal preference. In these cases, a
hybridization station may be utilized to improve mixing of
the hybridization solution over the array. The rate of
diffusion of a labeled nucleic acid in solution is actually
very low, and as such, some researchers prefer to use an
automated station that performs mixing of the solution.
In the case of Affymetrix GeneChip technology, a specific hybridization and fluidics station are required. The
hybridization station is simply a rotating incubator in
which the chips are placed. A bubble that is introduced
into the sealed array cartridge moves around during
While some microarray imagers such as the Perkin Elmer

ScanArray and GeneFocus DNAScope are confocal scanners, this is not a strict requirement. Confocal imaging
serves to eliminate extraneous signals, but reduces the
light gathering ability of the device. There are >10,000
commercial microarray scanners in the field capable of
reading standard glass microarrays. The leading scanner
makers include Agilent, Axon, Bio-Rad, GeneFocus, PerkinElmer, and other vendors. The laser scanner uses one or
more lasers with wavelengths appropriate to the fluorophores being used. The most commonly used fluorophores
for microarrays are cyanine 3 and cyanine 5 (or fluors with
equivalent spectra). Cyanine 3 has an absorbance maximum of 550 nm and emission maximum of 570 nm. There
are 2 main lasers used in scanners to excite this fluorphore:
Gre-Ne (green neon) gas lasers and Nd:YAG (neodymium
doped yttrium aluminum garnet) frequency doubled solidstate diode lasers. Cyanine 5 has an absorbance maximum
of 650 nm and an emission maximum of 670 nm. There are
two main lasers used in scanners to excite this fluorophore:
standard HeNe gas lasers and red diode lasers. Table 1
shows some of the characteristics of these two dyes, along
with two other popular dyes, Alexa 555 and Alexa 647,
which have spectra that are very similar to those of Cy3
and Cy5 respectively (Fig. 5).
Cyanine 3 and 5 have some important features that
make these dyes particularly suitable for use in microarray
analysis. The spectra of these dyes have little over lap and
can generally be separated from one another with little to
no cross-talk. In addition, these fluors have a somewhat
unique property in that they are brighter when dry than
when wet. Most fluorophores have the opposite behavior,
which is impractical for microarrays because the scanners
generally cannot handle wet preparations.
The other major class of microarray imager is a CCD
(charge coupled device) based system. In general, these
imagers use a white light source to excite the fluorophores.
The fluorescent light that is emitted is captured by the
CCD and converted into a digital image. Rather than
scanning the slide, a CCD based imager tiles together
several sections of the slide to create an image of the entire
surface. This tiling can create a stitching effect whereby
the seams of the images may not be completely smooth.
Table 1. Key Characteristics of the Most Commonly Used Fluorophores for Microarray Analysis
Fluorophore
Cy3
Cy5
Alexa555
Alexa647
Phycoerytherin
Excitation Max, nm
Emission Max, nm
Molar Extinction Coefficient
Molecular Weight
550
649
555
650
566
570
670
565
668
575
150,000
250,000
150,000
239,000
19,600,000
766
792
1,250
1,250
240,000
MICROARRAYS
550
600
650
wavelength (nm)
excitation/absorption ()
(- -) emission/fluorescence
500
Alexa Fluor 647/Cy5

(- -) emission/fluorescence
Alexa Fluor 555

Cy3
excitation/absorption ()
370
500
600
700
800
wavelength (nm)
Figure 5. Representative spectra of the fluors commonly used in spotted microarray experiments.
Alexa Fluor 555 and Cy3 are excited by green wavelengths of light whereas Alexa Fluor 647 and Cy5
are excited by red wavelengths of light. One green excited and one red excited fluor may be used at
the same time as there is little overlap in their excitation spectra.
This problem can be overcome with advanced lighting

systems and software.
Affymetrix arrays use a different labeling chemistry for
detection relying on the naturally occurring fluorescent
protein phycoerytherin. Phycoerythrin is a naturally
occurring pigment protein from light harvesting algae that
absorbs strongly at 566 nm and has an emission peak at
575 nm. It is a very bright fluorophore having a molar
extinction coefficient that is 80 times as high as the standard Cy3 and Cy5 molecules. The limitation of this molecule is that it is also 200 times larger, making the number
of molecules that can be incorporated per sequence
much less. As such, this molecule can only be applied to
the DNA posthybridization for fear that it would create
steric interference.
MICROARRAYS AS MEDICAL DEVICES
To date, microarrays have mostly found use in basic
research applications, and have yet to make a strong
impact on the diagnostic market. [During the preparation
of this text, Roche received FDA clearance for the first ever
array based diagnostic chip. The AmpliChip CYP450 based
on the Affymetrix platform was approved in January
of 2005 (see http://www.roche.com/med-cor-2005-01-12).]
Microarrays have indeed been used to study many diseases
including various cancers, cardiovascular disease, inflammatory disease, psychiatric disorders and infectious disease. This basic research will ultimately lead to the
identification of potential therapeutic markers for drugs
of for diagnostics. The potential of microarrays extends
beyond target discovery, however, and will eventually
impact on the way that medical care is performed.
Target Discovery
The use of microarrays in basic research laboratories
has often focused on target discovery. In these applications, microarrays are used to profile a particular
disease where disease tissues are compared to healthy

tissues either from the same patient or from a separate
test population. In such experiments, the goal is to find
genes that are differentially regulated (either up or down)
in the disease state compared to a healthy tissue. Such
genes are thought to be involved in the disease state or in
the cellular response to the disease. As such, these genes
are potential diagnostic markers and may also represent
drug targets.
Drug/Lead Discovery
Microarrays can also be used once the target has been
identified. It is possible to use microarrays to screen potential therapeutic compounds, for example, to determine
which candidates reverse the pattern of gene expression
that is indicative of disease. Microarrays have been even
more effective in looking at toxicity of lead compounds. One
of the leading contributors to failure of a pharmaceutical
compound is toxic or off target events. Microarrays have
proven useful in screening for the up-regulation in toxicity
related genes. In addition, it is possible to determine if the
compound creates other effects that while not toxic per se
could cause undesirable side effects from nonspecific interactions. Often toxicity models are tested in model organisms such as rats or dogs. Several toxicity specific arrays
have been developed that allow for profiling of genes in
these model systems rather than human cells.
Diagnostics and Prognostics
One of the more promising areas for microarrays to
have direct impact as a medical device is in the area of
diagnostics and prognostics. As mentioned under target
discovery, basic research has often strived to look for a
panel of genes that can be used as a molecular fingerprint
of a disease. There are numerous publications in
which researchers have attempted to use molecular
profiles to correlate to patient outcome, disease
state, tumor type, or any of several other factors. DNA
MICROARRAYS
microarrays are particularly well suited to this type of

analysis. Many complex diseases are multifactoral;
rather than a single prognostic or diagnostic marker being
present, it may be necessary to look at several genes at one
time. Microarrays allow for identification of a panel of
genes, which when looked at together may provide diagnostic or prognostic power. Although it has not become
common practice yet, there are examples of microarrays
being used to prescreen patients on the basis of a molecular
profile (14).
Other attempts are being made at using microarrays to
study infectious disease. Often times a patient may present
with a set of symptoms that could be indicative of several
different infectious agents. It is possible to prepare a
microarray that would identify the agent as well as to
subtype the bacterium or virus on the basis of pathogenicity. This particular application may prove very useful in
identifying not only the infectious agent, but also the best
course of treatment.
Pharmacogenomics and Theranostics
A concept that is gaining in popularity is pharmacogenomics or theranostics (15). Both of these terms refer to
the idea of tailoring a patients treatment or therapy on the
basis of their genetic makeup. Many pharmaceuticals on
the market have not known any potentially serious side
effects in a subset of patients. In addition, there are typically at least some patients that are nonresponders to a
particular treatment. These effects are often times the
result of the patients genetic make-up. Most of the work
in this area has focused on genotyping: looking at certain
variable regions of DNA and determining which variants
are present in people who have negative reactions or in
people who respond well to a treatment. It is hoped that in
the near future it will be possible to screen a patient and
determine which of a panel of drugs will be most beneficial.
Perhaps even more important, it will be possible to prevent
serious negative outcomes by avoiding treatment of a
patient that will have a poor reaction to a drug. Theranostics also involves monitoring a patient through a course of
treatment. It is possible that a patient can be screened
during treatment to ensure that the therapy is working as
expected. If a change occurs, the physician would be able to
alter the therapy to ensure that the disease is treated in the
most effective way possible.
SUMMARY
Microarrays provide a means to screen hundreds to thousands of biological analytes in parallel. These analytes can
be DNA, RNA, or protein. DNA microarrays allow for rapid
profiling of gene expression. While there are a few competing platforms that can be utilised, the basic principles are
the same: RNA from a biological sample is extracted,
labeled and applied to an array of DNA probes. Signals
generated from the array indicate which genes are active
and which are not. The ability to screen multiple tissues or
patients make microarrays particularly well suited to
uncovering the complex gene networks involved in disease.
While typically used in basic research applications for
371
target or marker discovery, the future will most likely

see microarrays used in diagnostic applications and for
tailoring medical treatment.
BIBLIOGRAPHY
1. Fodor SP, Rava RP, Huang XC, Pease AC, Holmes CP, Adams
CL. Multiplexed biochemical assays with biological chips.
Nature (London) 1993;364:555556.
2. Schena M, Shalon D, Davis RW, Brown PO. Quantitative
monitoring of gene expression patterns with a complementary DNA microarray. Science 1995;270:467470.
3. Cawley S, Bekiranov S, Ng HH, Kapranov P, Sekinger EA,
Kampa D, Piccolboni A, Sementchenko V, Cheng J,
Williams AJ, Wheeler R, Wong B, Drenkow J, Yamanaka
M, Patel S, Brubaker S, Tammana H, Helt G, Struhl K,
Gingeras TR. Unbiased mapping of transcription factor
binding sites along human chromosomes 21 and 22 points
to widespread regulation of noncoding rnas. Cell 2004;116:
499509.
4. Hegde P, Qi R, Abernathy K, Gay C, Dharap S, Gaspard R,
Hughes JE, Snesrud E, Lee N, Quackenbush J. A concise
guide to cdna microarray analysis. Biotechniques 2000;29:
548550, 552544, 556 passim.
5. Van Gelder RN, von Zastrow ME, Yool A, Dement WC,
Barchas JD, Eberwine JH. Amplified rna synthesized from
limited quantities of heterogeneous CDNA. Proc Natl Acad
Sci USA 1990;87:16631667.
6. Schena M. Microarray analysis. Hoboken: John Wiley &
Sons; 2003.
7. Wang Y, Wang X, Guo SW, Ghosh S. Conditions to ensure
competitive hybridization in two-color microarray: A theoretical and experimental analysis. Biotechniques 2002;32:
13421346.
8. Quackenbush J. Computational analysis of microarray data.
Nature Rev Genet 2001;2:418.
9. Singh-Gasson S, Green RD, Yue Y, Nelson C, Blattner
F, Sussman MR, Cerrina F. Maskless fabrication of lightdirected oligonucleotide microarrays using a digital micromirror array. Nat Biotechnol 1999;17:974978.
10. Gao X, LeProust E, Zhang H, Srivannavit O, Gulari E, Yu P,
Nishiguchi C, Xiang Q, Zhou X. A flexible light-directed
DNA chip synthesis gated by deprotection using solution
photogenerated acids. Nucleic Acids Res 2001;29:4744
4750.
11. LeProust E, Pellois JP, Yu P, Zhang H, Gao X, Srivannavit O,
Gulari E, Zhou X. Digital light-directed synthesis. A
microarray platform that permits rapid reaction optimization on a combinatorial basis. J Comb Chem 2000;2:349
354.
12. Hughes TR, Mao M, Jones AR, Burchard J, Marton MJ,
Shannon KW, Lefkowitz SM, Ziman M, Schelter JM, Meyer
MR, Kobayashi S, Davis C, Dai H, He YD, Stephaniants SB,
Cavet G, Walker WL, West A, Coffey E, Shoemaker DD,
Stoughton R, Blanchard AP, Friend SH, Linsley PS. Expression
profiling using microarrays fabricated by an ink-jet oligonucleotide synthesizer. Nat Biotechnol 2001;19:342347.
13. Hughes TR, Shoemaker DD. DNA microarrays for expression
profiling. Curr Opin Chem Biol 2001;5:2125.
14. Schubert CM. Microarray to be used as routine clinical
screen. Nat Med 2003;9:9.
15. Picard FJ, Bergeron MG. Rapid molecular theranostics in
infectious diseases. Drug Discov Today 2002;7:10921101.
See also DNA
REACTION.
SEQUENCE;
MICROBIOREACTORS;
POLYMERASE
CHAIN
372

PATRICIA L. DOLAN
JASON C. HARPER
SUSAN M. BROZIK
Sandia National Laboratories
Albuquerque, New Mexico
INTRODUCTION
Infectious diseases accounted for 2533% of the estimated
54 million deaths worldwide in 1988 (1), of which more
than half are attributed to tuberculosis, malaria, chronic
hepatitis B, diarrheal diseases, and human immunodeficiency virus/Acquired Immune Deficiency Syndrome HIV/
AIDS. The incidence of more than 30 diseases identified
since the mid-1970s continues to grow, which include HIV/
AIDS, liver disease due to hepatitis C virus, cholera, ticktransmitted Lyme disease, foodborne illness caused by E.
coli O157:H7 and Cyclospora, waterborne disease due to
Cryptosporidium, and the hantavirus pulmonary syndrome. Additionally, the first known cases of human influenza caused by the avian influenza virus, H5N1, were
identified in Hong Kong in 1997 (2).
Although death due to infectious diseases in the United
States remains low relative to that of noninfectious diseases, their occurrence is increasing. In 2000, the Federation of American Scientists reported that infectiousdisease-related death rates nearly doubled from 1980 to
170,000 annually (1). Many of these diseases, most recently
the West Nile virus, were introduced from outside the U.S.
borders by international travelers, immigrants, returning
U.S. military personnel, or imported animals and foodstuffs. Still, the most dangerous infectious microbes reside
within U.S. borders. Four million Americans are chronic
carriers of the hepatitis C virus, a significant cause of liver
cancer and cirrhosis. It is predicted that the death rate due
to hepatitis C virus infection may surpass that of HIV/
AIDS in the next five years. Influenza viruses are responsible for approximately 30,000 deaths annually. In addition, hospital-acquired infections are surging due to highly
virulent and resistant pathogens such as Staphylococcus
aureus.
The burden of identifying and treating infected individuals and controlling disease outbreaks generally lies with
physicians, hospitals, and first responders. Table 1 contains important characteristics for several of the more
common pathogenic microorganisms. As evidenced by this
noninclusive table, a wide variety of microorganisms exists
from which the specific diseasecausing microbe must be
identified. In addition, the number of cells or particles that
can provide an infectious dose is often extremely low. For
example, the infectious dose of E. coli O157:H7 is as low as
10 cells (3), which poses a significant challenge to healthcare professionals and first responders who must quickly
identify the infectious agent. Antimicrobial treatments
that attempt to neutralize all possible infectious pathogens
are often not possible or safe. Depending on the nature and
severity of the infection, a delay of only a few hours in
providing the proper therapy may lead to death.
Medical Microbiology
Medical microbiology is the discipline of science devoted to
identifying microbial agents that are responsible for infectious disease and elucidating the mechanism of interaction
between the microorganism and human host. Historically,
microbiologists have used plating, microscopy, cell culture,
and susceptibility tests to identify and study microorganisms. In the hospital and clinical diagnostic laboratory,
these, methods are still widely used and will be briefly
discussed in this article. The general procedure for isolation and identification of infectious and parasitic microbes
is (1) specimen collection and streaking onto culture plates
for production of isolated bacteria colonies, (2) staining and
microscopic analysis, (3) cell culture in various media, and
(4) antibiotic susceptibility testing.
Plating. Plating entails the streaking of a specimen
onto a solid nutrient media-filled Petri dish and incubation
at 3537 8C. Under these conditions, a single bacterium
divides and eventually produces a colony that is visible to
the eye (Fig. 1). A visible colony generally contains more
than 107 organisms. The colony morphology, color, time
required for growth, appropriate media, and other growth
conditions are used to characterize the microbe. The incubation time required for growth of a colony from a single
cell is dependent on the growth rate of the microorganism.
Fast-growing organisms such as E. coli, with a doubling
time of 30 minutes, would require approximately 13 hours
to produce a colony of 107 organisms. More typically, microorganisms require several days to a week to generate visible
colonies. The plated specimen is often a complex solution
such as blood, urine, feces, or sputum containing diverse
native flora in addition to the infectious microbe. Plating
serves as a method to isolate the infectious microbe as each
Figure 1. Colonies of E. coli grown on LB ampicillin and X-gal

IPTG agar media in a Petri dish.
373
small intestine
intestine
intestine
intestine
Salmonella typhi
Shigella spp.
Yersinia spp.
Vibrio spp.
nerves, muscle, skin,

intestine (infants)
intestine
skin lesion, large intestine
nerves, muscle
Clostridium botulinum
Clostridium difficile
Clostridium perfringens
Clostridium tetani
jaw, thorax, abdomen
meninges
small intestine
Neisseria meningitidis
Salmonella spp.
Anaerobes, Gram positive

Actinomyces spp.
large intestine
lower respiratory tract
genitourinary tract, eye
oropharynx
throat, skin, blood,

middle ear
lower respiratory tract,

skin lesions, intestine
skin, osteomyelitis, blood,
heart
upper respiratory track,

skin
lower respiratory tract
monocytes, leukocytes,
blood,
intestine, skin
Site(s)
Escherichia coli(O157:H7)
Legionella spp.
Neisseria gonorrhoeae
Bacteria, Gram negative

Bordetella pertussis
Streptococcus pyogenes
Staphylococcus aureus
Bacillus anthracis
Bacteria, Gram positive

Corynebacterium
diphtheriae
Streptococcus pneumoniae
Listeria monocytogenes
Organism
chronic abcesses,
draining sinuses
botulism, acute
bilateral cranial
nerve impairment
pseudomembranous colitis
gas gangrene, food poisoning
tetanus
shigellosis (enteritis); Bacillary

dysentery
enterocolitis, acute mesenteric
lymphadenitis
cholera, enteritis
typhoid fever
hemorrhagic colitis
Legionellosis, Pontiac fever
gonorrhoea, pelvic inflammatory
disease, septicemia, pharyngitis
meningitis
gastroenteritis
whooping cough
boils, furuncles, abscesses,

impetigo, osteomyelitis,
sepsis, toxic shock syndrome
pharyngitis, septicemia, erysipelas,
rheumatic fever, scarlet fever,
otitis media, foodborne illness
624 hours
14 days
321 days
variable; days or
months
1236 hours
496 hours
37 days
17 days
13 weeks
210 days
672 hours
28 days
210 days
27 days
620 days
13 days
variable; 410 days
few hours 7 days
13 days
variable; 370 days
pneumonia, meningitis
listeriosis, meningoencephalitis
anthrax
27 days
Incubation period
diphtheria
Disease(s)
Table 1. Characteristics of common pathogenic organisms
environmental
ingestion
direct contact
(noncommunicable)
direct contact (mouth),

airborne, fomites
direct contact, ingestion
(noncommunicable)
ingestion
ingestion
ingestion
ingestion
direct contact, droplet

spread, airborne
ingestion
airborne (noncommunicable)
direct contact (usually sexual),
neonatal contact (birth)
direct contact, droplet spread
ingestion

ingestion, direct contact,
neonatal
contact (birth),
transplacental
direct contact, airborne,
ingestion
direct contact, ingestion,
autoinfection,
spread, injection
Mode of transmission
unknown
105108 cells
potent toxin
unknown
unknown
106 cells
106 cells
unknown
10100,000
cells
1,000100,000
cells
10200 cells
10 cells
unknown
unknown
unknown
<1000 cells
8,00050,000
spores
102 106 cells
unknown
unknown,
<1000 cells
unknown
Infectious Dose
374
tract
tract, skin
tract, skin
tract, skin
liver
liver
skin
skin
upper respiratory tract
skin
skin
skin
Hepatitis C
Herpes simplex I
Herpes simplex II
Influenza
Measles
Rubella
VaricellaZoster
rubeola
German measles
chicken pox, shingles
herpes (vesicular lesions)

herpes (genital)
flu
hepatitis, type C
hepatitis, type B
hepatitis, type A
HIV/AIDS
aspergillosis
blastomycosis
coccidioidomycosis
histoplasmosis
oral thrush, intertrigo,

vulvovaginitis,
paronychia
meningitis, pneumonia
pulmonary, lymphadenitis
Tuberculosis
peritonitis, endometritis,
abscesses, septicemia
Disease(s)
813 days
1223 days
1317 days
710 days
212 days
14 days
610 weeks
24180 days
1050 days
6 months 7 years
variable; days to weeks

few weeks to months
14 weeks
317 days
unknown
variable; 25 days in
infants
unknown
412 weeks
unknown
Incubation period
(noncommunicable)
(noncommunicable)
(noncommunicable)
(noncommunicable)
direct and indirect contact

(bodily fluids), fomites
direct contact (contaminated
blood)
direct contact (saliva)
direct contact (usually sexual)
direct contact, droplet spread,
airborne
direct contact, droplet spread,
airborne
direct contact (sexual,

contact with blood
or bodily fluids)
ingestion
airborne
airborne
airborne
airborne
airborne
direct contact (sexual),


spread, airborne
ingestion, skin lesions
(noncommunicable)
endogenous
Mode of transmission
10 virus particles

1060 virus particles
unknown
102103
particles/mL blood
unknown
unknown
2800 virus particles
10103 virus
particles
unknown
unknown
unknown
unknown
unknown
10 spores
unknown
unknown
104107 cells
10 cells
unknown
Infectious Dose
Infectious Disease Information. (2005, April 29). Infectious disease information, NCID, CDC. [Online]. Centers for Disease Control and Prevention. http://www.cdc.gov/ncidod/diseases/index.htm [2005, August 21];
The Bad Bug Book. (2003, January 30). FDA/CFSAN Bad Bug Book: Introduction to Foodborne Pathogenic Microorganisms and Natural Toxins. [Online]. U.S. Food and Drug Administration Center for Food Safety
and Administration. www.cfsan.fda.gov/mow/intro.html [2005, August 21].
Infectious Agents MSDS Index. (2003, July 31). Index to Material Safety Data Sheets (MSDS) for Infectious Substances. [Online]. Public Health Agency of Canada. www.phac-aspc.gc.ca/msds-ftss/index.html [2005,
August 31].
respiratory
respiratory
respiratory
respiratory
progressive damage to
immune and other
organ systems
liver
lower
lower
lower
lower
meninges, lower
respiratory tract
mucous membranes, skin

laryngeal, meningeal
lymph nodes
mouth, respiratory track,

large intestine
Site(s)
Hepatitis B
Hepatitis A
Viruses
Acquired immunodeficiency
syndrome
Molds
Aspergillus spp.
Blastomyces dermatitidis
Coccidioides immitis
Histoplasma capsulatum
Cryptococcus neoformans
Yeasts
Candida albicans
Mycobacterium avium
complex
Bacteria, acidfast
Mycobacterium tuberculosis
Anaerobes, Gram negative

Bacteriodes spp.
Organism
Table 1. (Continued)
colony originates from a single cell and is therefore pure of

any other cell types. A colony can subsequently be used as a
pure sample for microscopy, cell culture, and other analytical tests. Additionally, as only viable cells divide, plating
can differentiate between dead microbes and those that are
viable and may be the source of infection.
Staining. Microscopic observation of microorganisms is
generally preceded by staining a specimen on a microscope
slide. The microbe response to various stains (gram-positive/negative, acid-fast, etc.), size, grouping (single, double,
chains), and morphology (bacillus, coccus, spirillum, pleomorphic) provide characteristic information helpful in
identifying the microorganism. However, several pathogenic species appear similar, or are indistinguishable,
under the microscope. For effective observation under a
microscope, at least 105 cells per milliliter of sample should
be present. A colony specimen usually meets this requirement, and preconcentration is generally necessary for
viewing a nonplated specimen. Still, very small cells can
be difficult to observe and may be overlooked. Finally,
microscopic observation usually cannot distinguish
between dead and live cells.
Cell Culture. Cell culture is used to ascertain the
biochemical properties of a microorganism. A single colony
is inoculated into a liquid media broth and incubated.
Incubation is usually performed near 37 8C with agitation
via shaking or gas sparging to facilitate gas transport into
the liquid media for uptake by the microbes. Signs of
microbial growth in liquid media include turbidity and
gas formation. Turbidity can be used as a simple and
nondestructive method to measure cell growth. An optical
density measurement provides the degree of light scattering at a particular wavelength through a given path length
of liquid media. Increasing cell density due to growth
usually increases the degree of light scattered. The measured optical density can, therefore, be directly related to
the total cell mass. A calibration curve for each bacterial
species is required as various sizes and shapes of different
microbes scatter light to varying extents.
Microbial growth in several different media is used to
determine a specific microbes biochemical and physiological characteristics. Definitive identification can require 20
or more media tests. Such tests often use selective media. A
media can be made selective by addition of chemicals that
inhibit microbe and native flora growth while allowing
growth of a specific organism. For example, ThayerMartin medium selectively isolates pathogenic Neisseria gonorrhea and Neisseria meningitides (4). The medium contains
vancomycin to inhibit growth of gram-positive bacteria,
anisomycin to inhibit fungi growth, colistin to inhibit most
gramnegative bacilli growth, and trimethoprim-sulfamethoxazole to inhibit Proteus growth. The Neisseria species are resistant to these inhibitors at the concentrations
present in the medium and grow freely.
Antibiotic Susceptibility Testing. Upon isolation and
identification of the infectious microbe, antibiotic susceptibility testing can be performed to identify antimicrobial
agents that inhibit growth. Additionally, the minimal
375
inhibitory concentration (MIC) is determined by exposing

bacteria in media broth to various concentrations of an
antimicrobial agent. The lowest antibiotic concentration
that inhibits growth is the MIC. A concentration of the
antibiotic in the blood at or above the MIC should successfully treat an infection.
Development
Plating, microscopy, cell culture, and susceptibility testing
techniques for identifying and treating infectious microorganisms have proven effective against a plethora of
pathogens, hence its continued use today. However, these
clinical microbiology methods have changed very little over
the past century, often require days to obtain confirmed
results, and cannot be used successfully to characterize
several significant infectious agents including the hepatitis
virus. However, with the recent and significant advances in
molecular biotechnology, two additional microbe identification methods have found wide use, immunoassay and
polymerase chain reaction.
Developed in 1959, the utility of the immunoassay was
not fully realized by the medical diagnostic community
until the late 1970s and early 1980s. The immunoassay
takes advantage of an immune system reaction, the highly
specific and strong binding of antibody to antigen. Antibodies are developed that specifically bind a given microorganism, chemical byproducts or proteins produced by a
given microorganism, or antibodies produced by the host
in response to infection caused by a given microorganism.
The developed antibodies are tagged with a reporter
molecule. Reporters can be radioisotopes, chemiluminescent or fluorescent molecules, or enzymes (i.e., alkaline
phosphatase, horseradish peroxidase) that can produce a
radiographic, colorimetric, or fluorescent signal. In the
presence of the antigen, the antibody will bind and will
remain bound through washes that remove unbound antibody. Detection of the reporter after washes indicates that
the antigen was present, as bound antibody was not
removed during washing. Although rapid, highly specific,
and sensitive, immunoassays cannot differentiate
between viable and dead cells and are limited to tests
for which antibodies can be developed. They also can be
affected by contaminants in the test specimen and do not
provide quantitative information regarding the number of
pathogenic agents present.
Serological assays are the most commonly used immunoassay in the medical laboratory and by the Centers for
Disease Control and Prevention (CDC). The mechanism of
Serodiagnosis entails binding of lab-developed antibodies
to antibodies produced in the host in response to a specific
infection. This indirect method of detecting infectious
agents allows identification of microbes that are currently
difficult or impossible to isolate and culture. For example,
because HIV-1 virus requires advanced containment facilities and is difficult to isolate and culture, it is serologically
diagnosed via detection of antibodies produced by the host
against the virus. Additionally, a method to effectively
isolate and culture hepatitis virus has not yet been devised.
Therefore, diagnosis of hepatitis virus infection is done
serologically. A lag phase of several weeks often exists
376
between onset of infection and production of antibodies by

the host against the microbial agent and, thus, possibly
leads to false negatives. False positives are also a concern
as antibodies produced by the host during a previous
infection may be present and detected.
Immunoassays were the primary diagnostic method
used for microbial detection by the CDC until the development of polymerase chain reaction (PCR) (5). PCR is a
technique that specifically amplifies DNA sequences (for
more information, see page 8). This technology has transformed molecular biology and genetics and has changed
diagnostic approaches to the identification, detection, and
characterization of infectious agents. With PCR, extremely
small quantities of DNA from a microorganism can be
amplified and detected. Detection of amplified DNA can
occur through gel electrophoresis or via genetic probes.
Based upon the highly specific binding between complimentary nucleobases of DNA and RNA, genetic probes are
nucleic acid sequences that bind to DNA or RNA unique to
a given microorganism. Genetic probes are marked with
radioisotopes, chemiluminescent or fluorescent molecules,
or enzymes and will produce a quantitative signal only
when the complimentary microorganism DNA or RNA is
present in the sample (for more information, see page 9).
Ou et al. (6) used PCR to amplify and detect HIV sequences
from seropositive individuals. Subsequently, PCR amplification and sequence analysis of HIV amplicons (amplified
DNA sequences) became the first use of comparative
nucleic acid sequence information in a disease outbreak
setting (7). Although PCR is very sensitive and sequence
analysis provides specific identification capability, these
technologies are expensive, time-consuming, labor-intensive, and require expertise in molecular biology. Consequently, use of PCR and genetic probes for identification of
microbes is common in research laboratories and academic
institutions, but, to date, is not widely used in hospital or
medical diagnostic laboratories.
To address rising national and worldwide public health
needs, it is desirable that a sensitive, specific, fast, and
simple-to-operate device be employed to detect infectious
agents. Microbial detection systems that attempt to meet
this need have been commercially available since the late
1970s and have progressed significantly with the molecular
biotechnology revolution. Still, microbial detection systems
face three major challenges: time, sensitivity, and specificity of analysis. Microbial testing and detection must be
rapid to allow adequate time for treating the infection and
be highly sensitive as a single pathogenic organism may be
infectious. Additionally, as a low number of pathogenic
microbes may be present in complex biological samples,
such as blood or urine, high specificity remains an essential
requirement. To tackle these problems, alternative nucleic
acid-based approaches have been integrated into userfriendly microbial detection systems that are commercially
available for diagnostic purposes.
Contemporary microbial detection systems or biosensors typically consist of a selective biorecognition molecule
connected to a transducer that converts a biochemical
interaction into a measurable signal. Recognition molecules include nucleic acids, antibodies, and peptides. Commonly used transducers include electrochemical, optical,
and piezoelectric. The following sections will discuss

numerous commercially available microbial detection systems used in clinical and field settings, including (1) nucleic
acid-based, optical technologies and systems; (2) fiberoptic, waveguide-based fluoroimmunoassay systems; (3)
a chip- and nanoparticle-based bio-barcode optical technology; (4) an electronic microchip-based technology; and (5)
an electronic nose microbial detector.
NUCLEIC ACID-BASED OPTICAL TECHNOLOGIES

Line Immunoprobe Assay (LIPA)
The line immunoprobe assay (LIPA) is a nucleic acid
recombinant immunoblotting assay (RIBA) (i.e., oligonucleotides that differentiate different genetic variants are
transferred onto a nitrocellulose membrane in a straight
line) (8,9). PCR is performed from the clinical sample using
primers that selectively amplify a DNA region containing
nucleotide differences. The amplicons are hybridized with
the immobilized oligonucleotides on the membrane, and an
enzyme-based colorimetric method is used to detect binding and positive reactivity. The nucleotide differences contained within the amplified sample DNA provide a unique
signature that differentiates target genotypes or mutant
microorganisms. These assays were among the first commercially available assays using nucleic acid hybridization
for diagnostic purposes.
COBAS AMPLICOR Analyzer
A second commercially available system using PCR technology is the COBAS AMPLICOR Analyzer (Roche Diagnostics; Rotkreuz, Switzerland). This system automates
amplification and detection of target DNA from infectious
agents by combining five instruments into one: a thermal
cycler, automatic pipettor, incubator, washer, and reader.
Amplified biotinylated products are captured on oligonucleotide-coated magnetic microparticles and detected colorimetrically with use of an avidin-horseradish peroxidase
(HRP) conjugate. The system can detect a broad range of
agents including Bacillus anthracis, Chlamydia trachomatis, Neiserria gonorrhea, Mycobacterium tuberculosis, cytomegalovirus, hepatitis B and hepatitis C viruses, and HIV
in clinical specimens including serum, urine, and sputum.
The manufacturer reports that more than 4000 COBAS
AMPLICOR Analyzers are currently used in clinical settings worldwide (10).
Real-Time PCR (RT-PCR)
A number of commercially available systems for the diagnosis of infectious diseases make use of a third nucleic acidbased approach (i.e., RT-PCR). As the name implies, RTPCR, pioneered by Applied Biosystems (Foster City, CA) in
the mid-1990s (11), amplifies and measures agent-specific
DNA as the reaction proceeds in real-time. It is used to
quantify the amount of agent-specific input DNA or cDNA
by correlating the amount of DNA with the time it takes to
detect a fluorescent signal. This technology uses fluorescent reporter probes (i.e., molecular beacons) that are
detected and quantitated at each cycle of the PCR. Molecular beacons are single-stranded, dual-labeled fluorogenic
DNA or RNA probes that form a stem loop structure. The
loop hybridizes to the target nucleic acid, whereas the stem
is end-labeled with a fluorophore at the 50 -end adjacent to a
quencher at the 30 -end. Fluorescence resonance energy
transfer (FRET) is the process by which energy from an
excited fluorophore (donor) is transferred to the adjacent
fluorophore (acceptor) at close proximity, resulting in the
quenching of fluorescence. Hybridization of the target
sequence to the loop separates fluorophore and quencher,
and the fluorescence is measured.
The GeneXpert System (Cepheid; Sunnyvale, CA) fully
automates and integrates sample preparation with the RTPCR detection processes. It uses microfluidics technology
integrated into disposable assay cartridges. The cartridges
contain all the specific reagents required to detect disease
organisms such as Bacillus anthracis, Chlamydia trachomatis, or foodborne pathogens. The system provides quantitative results from unprocessed clinical samples in 30
minutes or less and is capable is self-cleaning and decontamination before its next use. The GeneXpert module
forms the core of the Biohazard Detection System deployed
nationwide by the United States Postal Service for anthrax
testing in mail-sorting facilities (12). It is also used in
hospital laboratories, physician offices, and public health
clinics.
Idaho Technology (Salt Lake City, UT) manufactures an
automated, field-ready RT-PCR instrument, the R.A.P.I.D.
(ruggedized advanced pathogen identification device) system, which is based on the Light Cycler Instrument from
Roche Diagnostics (Alameda, CA). The R.A.P.I.D. is developed for military field hospitals and first responders in
harsh field environments. Amplification of DNA in realtime can be performed on environmental and blood samples. Idaho Technology claims a 15 minute set-up time and
a 20 minute PCR run for a total of 35 minutes using the
R.A.P.I.D. Pathogen Test Kit. This instrument is reported
to be very sensitive (i.e., Pseudomonas aeruginosa was
detected in blood culture samples at 10 cfu (colony-forming
units)/ml (13). (A colony-forming unit is a single viable cell
that forms a colony of identical cells when plated.) This
technology is well-established and has been in use worldwide since 1998.
The R.A.P.I.D. technology was recently put to the test at
the Prince Sultan Air Base in Saudi Arabia (14). Medical
personnel observed a clustering of diarrhea cases and
thought them to be due to influenza. However, testing of
patient samples with the R.A.P.I.D. identified the cause to
be foodborne Salmonella within hours of sample submission. Due to the prompt response by medical and services
personnel, the outbreak was limited to less than 3% of the
base population.
Nucleic Acid Sequence-Based Amplification (NASBA)
A fourth approach for nucleic acid-based detection of infectious organisms is nucleic acid sequence-based amplification (NASBA), a bioMe rieux, Inc. (Marcy-lEtoile, France)
proprietary isothermal amplification technology. This
method is based on specific amplification of RNA by the
377
NUCI. S NS4
ECL detection schematic diagram
PMT
magnetic
particle
capture
2
oligo RuRu(bpy)3 -oligo
RNA amplicon
electrode
Figure 2. NucliSens Reader ECL detection scheme (16).
simultaneous activity of three RNA-specific enzymes,

AMV-reverse transcriptase, T7 RNA polymerase, and
RNase H, generating single-stranded RNA as the endproduct (15). NASBA is an isothermal amplification procedure, carried out at 418C.
Two commercially available systems, NucliSens
Reader and NucliSens EasyQ Analyzer (bioMe rieux,
Inc.), make use of the NASBA technology. Although both
systems use NASBA for selective amplification of RNA,
as well as a bioMe rieux proprietary Boom silica-based
nucleic acid extraction method, the NucliSens Reader
relies on an electrochemiluminescence (ECL) detection
technology, and the NucliSens EasyQ uses fluorescent
detection by incorporating specific molecular beacons to
which amplicons hybridize. With this method, amplification and detection occur simultaneously in a
single tube.
The ECL-based NucliSens Reader employs a sandwich
hybridization method for the detection of amplified target
RNA (Fig. 2) (16). Two target-specific DNA probes are used:
a capture probe bound to magnetic beads and a detection
probe labeled with tris (2,20 -bipyridine) ruthenium (Ru).
Each of these probes bind to a different region of the target
RNA. After the hybridized sample is drawn into the ECL
flow cell and the beads are magnetically immobilized on the
electrode, a voltage is applied, and the resulting emitted
light is detected by a photomultiplier tube (PMT). According to the manufacturer, measurement of 50 reactions
takes approximately 50 minutes.
Real-time NASBA and fluorescent detection of targetbound molecular beacons are accomplished by the NucliSens Basic Kit and EasyQ Analyzer (Fig. 3) (17). This
technique is most often used to detect RNA viruses. Using
a multiplexed NASBA technique to detect four human
immunodeficiency virus type 1 (HIV-1) subtypes, DeBaar
et al. (18) reported an 89% correct subtype identification
relative to sequence analysis and a sensitivity of 92%. The
limit of detection was approximately 103 copies of HIV-1
RNA per reaction. Lanciotti and Kerst (19) conducted a
study comparing TaqMan RT-PCR (Applied Biosystems)
and standard reverse-transcription PCR (Roche Molecular Biochemicals) assays with NucliSens NASBA assays
378
Real-time Detection in NASBA

Target specific molecular beacons
sense RNA
RNase H
oligo PT
Oligo Pl
Reverse Transcriptase
RT
RNase H &
oligo P2
Reverse Transcriptase
anti-sense
RNA
T7 RNAP
T7 RNA Polymerase
Molecular beacon
hybridization
Figure 3. NucliSens EasyQ detection scheme

(17).
for detecting West Nile (WN) virus and St. Louis encephalitis virus. The ECL-based and molecular beacon-based
NASBA assays demonstrated equal or greater sensitivities and specificities than reverse-transcription PCR in
human cerebral spinal fluid. The NASBA-ECL assay for
WN virus was 10-fold more sensitive than either the TaqMan or NASBA-molecular beacon assay, detecting 0.01
pfu of WN virus. Moreover, the NASBA molecular beaconbased assay performed significantly faster than either
PCR procedures (i.e., a positive signal was detected within
1445 minutes).
Strand Displacement Amplification (SDA)
A fifth approach for nucleic acid-based detection of infectious organisms is strand displacement amplification
(SDA). SDA, first reported by Walker et al. in 1992 (20),
is an isothermal process that amplifies DNA or RNA using
a restriction enzyme and a DNA polymerase plus several
primers, without requiring temperature cycling. Available
since 1999, the BDProbeTecET System (Becton, Dickinson
& Co.; Franklin Lakes, NJ) couples the proprietary technology, SDA, and real-time fluorescent detection in a rapid
one-hour format. This high throughput, chip-based, closed
system was developed for detection of Chlamydia trachomatis and Neisseria gonorrhea in urine samples, endocervical swabs, and male urethral swabs in a one-hour assay
time. The complete system includes a sample preparation
module, a priming and warming heater unit, and an
amplification and fluorescence detection unit. The optical
system consists of a fiber-optic bundle with eight branches,
and the fluorescent detection reader monitors real-time
fluorescence by FRET. Emitted light passes through a
custom optical band-pass filter, is detected by a PMT,
and is analyzed by software.
Little et al. (1999) (21) reported a sensitivity of 1015 N.
gonorrhea cells or C. trachomatis elementary bodies.
Akduman et al. (22) reported that out of 3544
urine samples tested, 152 were positive using the BDPro-
beTecET System, and 130 were positive by standard culture techniques resulting in a sensitivity of 99.2% and a
specificity of 99.3%.
FIBER-OPTIC FLUOROIMMUNOASSAY SYSTEMS
Analyte 2000 and RAPTOR
The Analyte 2000 and its sister field model, RAPTOR
(Research International; Monroe, WA), detection systems
use a fiber-optic, waveguide-based sandwich fluoroimmunoassay for the near real-time detection of pathogens in a
variety of raw fluid samples (23). Optical fibers are long,
thin strands of either glass or plastic that can transmit
light over long distances. In the RAPTOR, a monolayer of
capture antibodies are immobilized on the surface of a
cylindrical waveguide (Fig. 4)(23). The waveguide is incubated with a clinical sample for three to five minutes,
washed, and re-incubated with a fluorophore-labeled antibody to form an antibody/antigen/labeled-antibody sandwich. Excitation light, injected into the waveguide,
creates an evanescent wave electric field in the fluid and
generates an optical emission from the antibody-antigen
complexes. The fluorescent signals are then monitored by a
photodetector.
Using the Analyte 2000, the detection limits for Bacillus anthracis (vegetative cells) was reported as 30 cfu/ml
in water, and for the avirulent strain of B. anthracis (i.e.,
Sterne strain), 100 cfu/ml in whole blood. For spores, the
detection limit was 5 104/ml (23). The infectious dose of
B. anthracis in a healthy individual requires inhalation of
about 8,00050,000 spores (24). This number is reduced
in more vulnerable individuals, such as the elderly or
those with respiratory problems. Vaccinia virus (a surrogate of the Smallpox virus) from throat swabs was
detected at 2.1 104 pfu (plaque-forming units, the viral
equivalent of bacterial colonies)/ml (25). The infectious
dose of smallpox is thought to be low (i.e., 10100 organisms) (26).
379
Figure 4. Optical and biomolecular processes of

RAPTOR technology (23).
NANOPARTICLE-BASED BIO-BARCODE TECHNOLOGY

Verigene System
The Verigene System (Nanosphere; Northbrook, IL) is an
automated device for the chip-based detection of proteins
and nucleic acids using an innovative gold nanoparticlebased bio-barcode technology. For proteins, the assay uses
two types of probes (Fig. 5 (27): (1) magnetic microparticles
(MMPs) functionalized with monoclonal antibodies (mAbs)
specific for a target antigen and (2) gold nanoparticles (NP)
functionalized with polyclonal antibodies specific for the
same target and DNA oligonucleotides (the bio-barcodes)
with a sequence that is a unique identification tag for the
target. The Au nanoparticles and the MMPs sandwich
the target, generating a complex with a large ratio of
barcode DNA to protein target. A magnetic field is applied,
allowing the separation of all the MMP/target/NP
complexes from the reaction mixture. After a wash to
dehybridize the barcode DNA from the nanoparticles,

another magnetic field removes the NPs, leaving only
the barcode DNA. Detection and identification of the barcodes occurs next through a PCR-less process of amplification. Chip-immobilized capture DNA, complementary
with half of the target barcode DNA sequence, is used to
bind the barcode DNA. Then, gold nanoparticles, functionalized with oligonucleotides that are complementary to
the other half of the barcode DNA, are hybridized to the
captured barcode strands. The signal is amplified by the
catalytic electrodeposition of Ag onto the Au nanoparticles, and the results are recorded with the Verigene ID
system, which measures scattered light intensity from
each barcode/Au/Ag complex.
Like protein detection, DNA detection via the nanoparticle bio-barcode approach uses two types of probes
(Fig. 6)(28): (1) magnetic microparticles functionalized
with oligonucleotides that are complementary to one-half
Figure 5. Prostate-specific antigen

(PSA) detection and barcode DNA
amplification and identification (27).
380
of a target sequence and (2) gold nanoparticles functionalized with two types of oligonucleotides, one that is complementary to the other half of the target sequence and one
that is complementary to a barcode sequence that is a
unique identification tag for the target sequence. The assay
proceeds in the same manner as with protein targets, with
the analysis also accomplished by the scanometric method
with a Verigene ID system.
The nanoparticle-based bio-barcode approach is
reported to provide a sensitivity of 500 zeptomolar,
approximately 10 target DNA strands in a 30 ml sample
(27). Prostate-specific antigen was detected at 30 attomolar
levels with this method, and PCR on the DNA barcodes
boosted sensitivity to 3 attomolar (28). The entire assay can
be carried out in 34 h.
MICROCHIP TECHNOLOGY
NanoChip System
Figure 6. The DNA-bio-barcode assay. (a) Nanoparticle and

magnetic microparticle probe preparation. (b) Nanoparticlebased PCR-less DNA amplification scheme (28).
The NanoChip System (Nanogen; San Diego, CA) is an

electronic microarray device based on the electrophoretic
transport of proteins and nucleic acids on a microchip to
specific sites where traditional immunoassays or nucleic
acid hybridization reactions occur (Fig. 7) (29). The electronic microchip is a planar array of microelectrodes that
electrophoretically transport-charged biomolecules to any
individually-electrically-addressed microsite on the surface of the device. Each microsite has an agarose-streptavidin permeation layer coated on top of a platinum
microelectrode to bind biotinylated capture molecules.
The microchips are referred to as active electronic microchips because electric fields are generated for the
purpose of transporting biomolecules to and from specific
a
Assay microlocations
Auxiliary electrodes
b
Analyte
Immobilized
capture antibody
+ Electrode
Figure 7. Active electronic microchip technology. (a) Basic

chip layout; (b) Cross-section of microchip for electrophoresis of proteins (29).
Permeation layer
Electrode
Si O2
Si Substrate
Assay microlocation
microsites in a process the manufacturer terms electronic

addressing (29). As the electric field is generated only in
the immediate vicinity of the electrodes, not affecting the
solution in other parts of the device, each microsite is an
independent assay site allowing for the detection of multiple analytes. Also, the generated electric fields can be used
to selectively dehybridize nonspecifically bound analytes
from assay sites, which greatly improves the selectivity of
the assay. When the biotinylated capture probes are
attached to the array, fluorescently labeled analytes are
introduced, and further electrical adjustments are made to
direct the analytes to concentrate at the microsites for
rapid hybridization or antibody-antigen interactions. The
fluorescent signal is monitored in a laserinduced fluorescence scanner, and the analytes are identified based on the
microlocation of the fluorescence.
Nanogen researchers performed a diagnostic immunoassay for two fluorescently labeled toxins simultaneously, staphylococcal enterotoxin B (SEB) and cholera
toxin B. They reported a sensitivity of better than 20 nM
concentrations of toxins (29). High specificity was also
demonstrated by low nonspecific binding and cross-binding. This assay took 6 minutes to perform, 1 minute for
electronic addressing to bind analytes and 5 minutes for
washing to reduce nonspecific binding.
More recently, Nanogen researchers reported on
an integrated stacked microlaboratory for performing
automated electric field-driven immunoassays and DNA
hybridization assays (30). This device is composed of a
CMOS-based electronic microarray chip, a dielectrophoresis microchip, and several modules for DNA sample
preparation, strand displacement amplification, and
hybridization. E. coli bacteria and Alexa-labeled staphylococcal enterotoxin B were detected in the device with
specific-to-nonspecific signal ratios of 4.2:1 and 3.0:1,
respectively. Identification of the Shiga-like toxin gene
from E. coli was accomplished in a 2.5 h comprehensive
protocol including the dielectrophoretic concentration of
intact bacteria, DNA amplification, electronic DNA hybridization to fluorescently-labeled probes, and detection
with a fluorescent microscope. This experiment used bacteria cell suspensions of 109 cells/ml with a specific-tononspecific signal ratio of 22.5:1, showing outstanding
specificity.
381
ELECTRONIC NOSE
Osmetech Microbial Analyzer
An electronic nose is a device that consists of an array of gas
sensors with different selectivity patterns, a signal collecting unit, and data analysis by pattern recognition software.
When microorganisms grow and metabolize, they emit
volatile organic compounds and gases that can be monitored by a biosensor array. The Osmetech Microbial Analyzer (OMA; Osmetech; London, UK) is an automated
headspace analyzer using arrays of organic conducting
polymers as sensors. The device samples the headspace
above the surface of the specimen and detects volatile
compounds with an array of up to 48 conducting polymer
sensors. Each polymer has unique adsorptive properties,
and, once adsorbed, the volatile components modulate the
conduction mechanism of the polymer resulting in reversible changes in resistance (Fig. 8) (31). The signal is
measured as a percentage change of the original resistance
of the polymer. Multivariate data algorithms are used to
compare the responses and establish a diagnosis.
When 534 clinical urine samples were analyzed by the
OMA, 22.5% had significant bacteriuria (i.e., >105 cfu/ml),
resulting in a sensitivity of 84% and a specificity of 88%
relative to standard culture methods (32). Alhough less
than optimal, this device shows promise for automated,
rapid screening. The companys second FDA approval for
detection of bacterial vaginosis was secured in January
2003. Clinical trials with more refined versions of the
instrument are in progress. Although electronic nose technology is still in its infancy, it clearly has the potential for
providing rapid, sensitive, and simultaneous detection of
different strains of bacteria.
FUTURE TRENDS IN MICROBIAL DETECTION SYSTEMS
In the development of the microbial detection systems
mentioned, researchers have begun to focus on building
integrated devices that combine a pre or post-processing
step such as PCR-based amplification, with post-derivatization (fluorescent labeling) and detection. Microarray
technologies are being developed to overcome limitations
of sample volume and high throughput analysis. In the
Figure 8. Osmetch Microbial Analyzer detection

technology (31).
382
near future, biomedical science can realize the integration

of all laboratory equipment used in molecular biology on a
chip-based platform arrayed to detect large numbers of
pathogens in a high throughput, portable device. Biological
Micro-Electro-Mechanical systems (BioMEMS), also
referred to as lab-on-a-chip and micro total analytical
systems (m-TAS), is an area rapidly advancing due to the
integration of micro and nanotechnology with biotechnology. Current reviews for this technology are abundant in
the literature (33,34). Microfluidic-based devices have been
on the market since 1999, but much work is still underway
to build modular-type systems with complete integration of
sample collection, concentration, pre and post-processing
steps, separation, selective capture, viability detection,
lysing, and protein and DNA analysis. Development of
such systems in a high throughput fashion capable of
detecting and discriminating between hundreds of pathogenic agents would impact not only medical diagnostics but
homeland security and public health, including home monitoring, medicine, and veterinary diagnostics. As the field
moves toward lab-on-a-chip systems, cost, limited sample
throughput, ease-of-use, and limited waste production
(reagentless systems) will be considered in design strategies. The second progression toward advanced microbial
detection systems will be the incorporation of nanotechnology. Current nanotechnologies such as quantum dots,
nanoparticles, and synthetic nanopores are already being
incorporated into current chip-based diagnostic systems.
CellTracks technology (Immunicon Corporation, Huntingdon Valley, PA) has developed magnetic nanoparticles
called ferrofluids, which consist of a magnetic core encompassed by a polymer coating tagged with antibodies for
whole cell and pathogen detection. Up-Converting Phosphor Technology (UPT), by OraSure Technologies, Inc.,
makes use of proprietary ceramic nanoparticles for DNA
detection. These particles have been shown to be a 1000
times more sensitive than fluorescent technologies.
Finally, a trend exists to build detection systems from
the bottom up rather than the top down. Small building
blocks such as protein motors are being designed to move
cargo including peptides and antibody fragments as a
method of patterning arrays. Switchable materials such
as poly-n-isopropylacrylamide (PNIPAM) are used to pattern antibodies, capture proteins, and move fluids, replacing mechanical components of BioMEMS systems.
PNIPAM has a thermally activated lower critical solubility
temperature (LCST) of 32 8C. At temperatures below the
LCST, the polymer swells in water to create a hydrophilic
surface that resists protein adsorption. Above the LCST,
the polymer collapses to form a hydrophobic surface that
promotes protein adsorption. Whether the bottom up
approach based solely on nanomaterials will hold in the
long run remains to be seen. However, it is clear that
nanotechnology that complements and extends current
MEMS detection methods will revolutionize the field of
medical diagnostics. Early examples of lab-on-a-chip technologies integrating nanotechnologies already exists,
which address the current limitations of detection systems.
The approach is toward development of portable microsystems that are reagentless; handle small sample size;
eliminate the need of labels and probes; are specific, sensi-
tive, and high throughput; perform multiple functions

from sample concentration to final detection; and are easy
to use.
Briefly, some of these technologies include cantilever
arrays, which operate by a slight bending of the cantilever
beam at the nanoscale level upon analyte binding. Protiveris, Inc. (Rockville, MD) is developing microcantilever
arrays for combined detection of DNA and protein. Capture
molecules are attached to the beams and, as samples moves
across the device, binding of a target molecule results in nm
bending of the beam. These devices can be integrated into
microfluidic systems, require no labels or reagents, and are
very sensitive and specific. Nanowires, nanoneedles, and
nanoelectrode arrays are additional technologies that can
detect multiple analytes simultaneously. Electronic signals can be averaged over thousands of electrodes eliminating the need for PCR amplification, and no reagents are
required. These devices are coated with selective molecular
recognition molecules and change in conductance occurs
during a binding/recognition event. Aside from integration
into lab-on-a-chip systems, these technologies have applications in in vivo medical diagnostics.
Over the next few years, nanotechnologies will continue
to evolve and become integrated into chip-based microsytems for detection, diagnostics, and drug delivery. Later, in
perhaps 20 30 years, the introduction of nanomachines for
in vivo diagnostics and treatment may well emerge, changing the current way of conducting medical and healthcare practice.
BIBLIOGRAPHY
1. Gannon JC. The Global Infectious Disease Threat and Its
Implications for the United States. [Online]. Federation of
American Scientists. http://www.fas.org/irp/threat/nie9917d.htm.
2. Fauci AS. Global Health: The United States Response to
Infectious Diseases, Testimony before the U.S. Senate Labor
and Human Resources Subcommittee on Public Health and
Safety. [Online]. National Institute of Allergy and Infectious
Diseases, National Institutes of Health. http://www3.niaid.nih.gov/about/directors/congress/1998/0303/default.htm.
3. Ivnitski D, Abdel-Hamid I, Atanasov P, Wilkins E. Biosensors
for detection of pathogenic bacteria. Biosens Bioelectron
1999;14:599624.
4. Washington JA. Principles of Diagnosis. [Online]. University
of Texas Medical Brach. http://gsbs.utmb.edu/microbook/
ch010.htm.
5. Saiki RK, Gelfand DH, Stoffel S, Scharf SJ, Higuchi R, Horn
GT, Mullis KB, Erlich HA. Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase.
Science 1988;239:487491.
6. Ou CY, Kwok S, Mitchell SW, Mack DH, Sninsky JJ, Krebs
JW, Feorino P, Warfield D, Schochetman G. DNA amplification for direct detection of HIV-1 in DNA of peripheral-blood
mononuclear-cells. Science 1988;239:295297.
7. Ou CY, Ciesielski CA, Myers G, Bandea CE, Luo CC, Korber
BTM, Mullins JI, Schochetman G, Berkelman RL, Economou
AN, Witte JJ, Furman LJ, Satten GA, Macinnes KA, Curran
JW, Jaffee HW. Molecular epidemiology of HIV transmission
in a dental practice. Science 1992;256:11651171.
8. Stuyver L, Van Geyt C, De Gendt S, Van Reybroeck G, Zoulim
F, Leroux-Roels G, Rossau R. Line probe assay for monitoring
MICROBIOREACTORS
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
drug resistance in hepatitis B virus-infected patients during

antiviral therapy. J Clin Microbiol 2000;38:702707.
Stuyver L, Wyseur A, Vanarnhem W, Lunel F, Laurentpuig
P, Pawlotsky JMM, Kleter B, Bassit L, Nkengasong J, Vandoorn LJ, Maertens G. Hepatitis C virus genotyping by
means of 50 -UR/core line probe assays and molecular analysis
of untypeable samples. Virus Res 1995;38:137157.
Roche Diagnostics - Roche Molecular Diagnostics - Products.
Automated PCR Systems. [Online]. Roche Molecular
Diagnostics. http://www.rochediagnostics. com/ba_rmd/
rmd_ products_automated_pcr_systems23.html.
Heid CA, Stevens J, Livak KJ, Williams PM. Real time
quantitative PCR. Genome Res 1996;6:986994.
Cepheid Products. (No date). GeneXpert Technology, System Overview. [Online]. Cepheid. http://www.cepheid.com/
Sites/cepheid/content.cfm?id164.
Jaffe RI, Lane JD, Bates CW. Real-time identification of
Psuedomonas aeruginosa direct from clinical samples using
a rapid extraction method and polymerase chain reaction
(PCR). J Clin Lab Analysis 2001;15:131137.
Idaho Technology - R.A.P.I.D. Successes using the R.A.P.I.D.
[Online]. Idaho Technology. http://www.idahotechnology.com/rapid/success.htm.
NucliSens Key Technologies. (2005). bioMerieux Clinical
Microbiology Products. [Online]. bioMerieux, Inc. http://
www.biomerieuxusa. com/clinical/nucleicacid/technology. htm.
NucliSens Reader. (2005). bioMerieux Clinical Microbiology
Products. [Online]. bioMerieux, Inc. http://www.biomerieuxusa.com/clinical/nucleicacid/reader.htm.
NucliSens EasyQ. (2005). bioMerieux Clinical Microbiology
Products. [Online]. bioMerieux, Inc. http://www.biomerieuxusa.com/clinical/nucleicacid/easyq/easyq_technology.htm.
DeBaar MP, Timmermans EC, Bakker M, Rooij E, van
Gemen B, Goudsmit J. One-tube real-time isothermal amplification assay to identify and distinguish human immunodeficiency virus type 1 subtypes A, B, and C and circulating
recombinant forms AE and AG. J Clin Microbiol
2001;39:18951902.
Lanciotti RS, Kerst AJ. Nucleic acid sequence-based amplification assays for rapid detection of West Nile and St. Louis
encephalitis viruses. J Clin Microbiol 2001;39:45064513.
Walker GT, Little MC, Nadeau JG, Shank DD. Isothermal in
vitro amplification of DNA by a restriction enzyme/DNA
polymerase system. Proc Natl Acad Sci USA 1992;89:392
396.
Little MC, Andrews J, Moore R, Bustos S, Jones L, Embres C,
Durmowicz G, Harris J, Berger D, Yanson K, Rostkowski C,
Yursis D, Price J, Fort T, Walters A, Collis M, Llorin O, Wood
J, Failing F, OKeefe C, Scrivens B, Pope B, Hansen T, Marino
K, Williams K, Boenisch M. Strand displacement amplification and homogeneous real-time detection incorporated in a
second-generation DNA probe system, BDProbeTecET. Clin
Chem 1999;45:777784.
Akduman D, Ehret M, Messina K, Ragsdale S, Judson FN.
Evaluation of a strand displacement amplification assay
(BD ProbeTec-SDA) for detection of Neisseria gonorrhoeae
in urine specimens. J Clin Microbiol 2002;40:281
283.
RAPTOR. RAPTOR, Portable, Multianalyte Bioassay System. [Online]. Research International. http://www.resrchintl.com/raptor.html.
Bacillus anthracis. (June 2, 2003). Bacillus anthracis
(anthrax). [Online]. http://microbes.historique.net/anthracis.
html.
Donaldson KA, Kramer MF, Lim DV. A rapid detection
method for Vaccinia virus, the surrogate for smallpox virus.
Biosens Bioelectron 2004;20:322327.
383
26. Franz DR, Jahrling PB, Friedlander AM, McClain DJ, Hoover
DL, Bryne WR, Pavlin JA, Christopher GW, Eitzer EM, Jr..
Clinical recognition and management of patients exposed to
biological warfare agents. JAMA 1997;278:399411.
27. Nam J-M, Thaxton CS, Mirkin CA. Nanoparticle-based biobarcodes for the ultrasensitive detection of proteins. Science
2003;301:18841886.
28. Nam J-M, Stoeva SI, Mirkin CA. Bio-bar-code-based DNA
detection with PCR-like sensitivity. J Am Chem Soc
2004;126:59325933.
29. Ewalt KL, Haigis RW, Rooney R, Ackley D, Krihak M.
Detection of biological toxins on an active electronic microchip. Anal Biochem 2001;289:162172.
30. Yang JM, Bell J, Huang Y, Tirado M, Thomas D, Forster AH,
Haigis RW, Swanson PD, Wallace RB, Martinsons B, Krihak
M. An integrated, stacked microlaboratory for biological
agent detection with DNA and immunoassays. Biosens Bioelectron 2002;17:605618.
31. Osmetech. (2005). eNose Technology. [Online]. Osmetech.
http://www.osmetech.plc.uk/enose.htm.
32. Aathithan S, Plant JC, Chaudry AN, French GL. Diagnosis of
bacteriuria by detection of volatile organic compounds in
urine using an automated headspace analyzer with multiple
conducting polymer sensors. J Clin Microbiol 2001;39:2590
2593.
33. Bashir R. BioMEMS: State-of-the-art in detection, opportunities and prospects. Adv Drug Delivery Rev 2004;56:1565
1586.
34. Lee SJ, Lee SY. Micro total analysis system (m-TAS) in
biotechnology. Appl Microbiol Biotechnol 2004;64:289299.
See also COLORIMETRY;
COMPUTER-ASSISTED DETECTION AND DIAGNOSIS;
COMPUTERS IN THE BIOMEDICAL LABORATORY; FLUORESCENCE MEASUREMENTS; SAFETY PROGRAM, HOSPITAL.
MICROBIOREACTORS
XIAOYUE ZHU
TOMMASO BERSANO-BEGEY
YOKO KAMOTANI
SHUICHI TAKAYAMA
University of Michigan
Ann Arbor, Michigan
INTRODUCTION
This article defines microbioreactors as micrometer scale
reaction vessels in which biological reactions are performed. Whereas large-scale bioreactors mainly focus on
efficiently producing desired end products, microbioreactors have additional targeted applications such as studying
cellular processes under simulated physiological microenvironments, and functioning as portable cellular biosensors or implantable devices inside the body to restore tissue
functions. Increasing needs in developing personalized
cell-based therapies and medicines together with rapid
advances in micro- and nanotechnologies ensure that the
field of microbioreactors will continue to flourish. Although
most microbioreactors are still in their infancy, relatively
simple compared to their macroscopic counterparts, and
often not fully integrated or packaged into compact selfcontained platforms, they already have started to have an
384
MICROBIOREACTORS
Figure 1. What are bioreactors and microbioreactors? Part

(a) shows a generalized bioreactor application, feeding cells to
produce complex biomolecules for pharmaceutical applications.
Part (b) shows the main components of both bioreactors and
microbioreactors: a reaction chamber containing main reagents
and cells, sensors (for temperature, flow rate, pH, oxygen, glucose,
etc.), actuators (pumps, valves, heaters, mixers, filters), and
controllers to regulate actuators based on sensor readings.
impact in pharmaceutical, medical, clinical, and biological

fields (Fig. 1).
Microbioreactors enable low cost high throughput investigation of bioreactions in ways not possible with macroscopic bioreactors. For example, the development of a
biopharmaceutical requires extensive optimization processes prior to scale-up for industrial manufacture: The
bacterial strain or mammalian cell that produces the highest yield of a particular product is constructed or identified;
cultural parameters (e.g., temperature, pH, oxygen concentration, and media composition) are systematically
evaluated to maximize cell growth and product formation;
the design of the reactors themselves are adjusted iteratively as the development process gradually scale-up from
benchtop to production scale. With microfabrication techniques, hundreds and thousands of small reaction chambers can be produced simultaneously and operated in
parallel allowing vast combinations of parameters to be
screened concurrently in short periods of times (Fig. 2a).
The consumption of reagents and resources are reduced
dramatically because for a given reaction chamber geometry, every 10-fold reduction in linear dimension would lead
to a 1000 time reduction in its volume.
To enhance the quality of information obtained from
cellular studies, rather than simply increasing throughput,
some microbioreactors are designed to simulate physiological cellular microenvironments. Using microbioreactors
for studying cellular physiology makes intuitive sense
when one considers that much of the bioreactions within
living organisms occur at the microscale. For example,
capillary blood vessels, lung small airways, livers sinusoids, kidney nephrons, and reproductive tracts are all
networks of small sacs, ducts, and tubes. Cells in these
and other similar systems are constantly perfused with
nutrients and oxygen and exposed to shear stresses and
gradients of chemicals (13). Conventional in vitro cell
culture studies, such as culture dishes or 96 well plates,
often fail to present many of these dynamic physiological
parameters. Microbioreactors, however, can be designed to
simulate many such physical and chemical conditions that
cells experience inside the body. In the body, different
tissues and cell types also interact and communicate with

each other. Microbioreactors can be designed to network
multiple reaction chambers together to capture the complexity of living organisms and used as animal and human
surrogates in pharmacokinetic and toxicology studies.
Microbioreactors can also work as cell-based biosensors,
where effects on cell behaviors serve as readouts for selective and sensitive detection (Fig. 2c).
Some microbioreactors aim to continuously produce and
deliver small quantities of biopharmaceuticals (e.g., insulin for regulation of blood sugar) inside the body (Fig. 2e).
Implantable devices that continuously produce drugs
based on physiological demands would eliminate the need
for repeated injections and blood tests, and allow for delivery of stable, safe, and effective doses that resemble physiological concentration profiles. Such devices contain cells
that use nutrients from the body to produce and secrete
drugs (4).
Besides cell-based microbioreactors, enzymatic microbioreactors are also useful for detection and analysis.
Enzyme-linked immunosorbent assays (ELISA), for example, are useful in high sensitivity detection and analysis of
a wide variety of proteins and chemicals related to pollution, disease, and basic biology. Microscale systems often
require shorter incubation time due to the short distances
that analytes and reagents have to diffuse. Microbioreactors are also more portable and thus suitable for field and
point-of-care use.
The development, integration, and packaging of microbioreactors present many challenges, as well as unique
opportunities. Because of scaling issues and fabrication
limitations, functional microscopic devices often require
totally new designs instead of simply shrinking their
macroscopic counterparts. Thus microscale pumps, valves,
and mixers not only look different from their macroscopic
counterparts, but may also operate with different mechanisms. Microscale sensors, another crucial component of
microbioreactors, is an active area of research that has
yielded a variety of useful systems based on optical, electrochemical, ultrasonic, and mechanical detection
schemes. Although many microscale components have
been developed to date, functional microbioreactors that
integrate multiple pumping, valving, mixing, sensing, and
control features are still relatively uncommon. Microbioreactors generally have fewer components integrated into
their systems compared to their larger counterparts.
The organization of the remainder of this article is as
follows. First, the principles that govern microscale reactions and microbioreactor operation are discussed. Second,
elements of microbioreactor components are described,
along with a brief description of the microfabrication processes that can be used to create them. Finally, highlights
of state-of-the-art microbioreactors are given with focuses
on production optimization, clinical treatment, toxicology
testing, development of implantable systems, and basic cell
biology studies. Because of space limitations, this article is
representative rather than comprehensive. We also focus
mainly on cell-based bioreactors rather than enzyme-based
ones. Interested readers are referred to reviews on immobilized microfluidic enzymatic reactors (5). We also exclude
important bioreactor categories that are not directly
MICROBIOREACTORS
385
Figure 2. Advantages and new applications of microbioreactors over large scale bioreactors:
(a) microscale parallel cell culture and testing in a microchip configuration can be used to test
and compare hundreds of conditions and parameters simultaneously, to optimize a reaction or
screen potential treatments. (b) Microdevices containing interconnected chambers and
microcultured cells from different tissues can be used to test drug effects on entire organisms,
taking into account systemic impact of organs such as liver on drug metabolism, possibly replacing
some stages of animal and human testing. (c) Microbioreactors can be used as detectors for bioactive
compounds (e.g., toxins, endocrine disruptors, drugs) or biodisruptive conditions (e.g., radiation)
with the advantage that all the steps of a test can be integrated in a single device that requires only
trace amounts of samples and reagents. Cell-based sensors may be able to detect unknown reagents
that affect living organisms. (d) Microfabrication can also provide microbioreactors with more
physiologically accurate microenvironments, thus making in vitro testing more reliable and closer to
in vivo conditions: for example, in actual tissues, each cell is held in its three-dimensional (3D) shape
by tension through connected cytoplasmic fibers (the cytoskeletons), and cells have many surface
interactions with other surface microfeatures and other cells. In contrast, in conventional cell
culture that cannot microfabricate these features, cells configuration, and environment conditions,
such as the amount of surface contacts, are drastically different. (e) Implantable devices containing
cells such as insulin-producing pancreatic islets can be used as an improved treatment that
continuously produce insulin based on the bodys minute-by-minute needs, eliminating the need
for periodical blood tests and consequent injections of insulin mass-produced in conventional
bioreactors.
related to medicine, such as those used for food testing,

plant cell culture, and wastewater treatment.
MICROBIOREACTOR DESIGN PRINCIPLES
Basic principles of bioreactor operations, such as thermodynamics, kinetics, mass transfer, sterilization, and struc-
tural considerations, are similar for both macro- and

microbioreactors. The implementation of these principles
into actual practice, however, can differ greatly.
Scaling Effects
At first glance, development of microbioreactors may seem
to be a matter of simply shrinking macroscopic components
386
MICROBIOREACTORS
Figure 3. Microscale physics useful

for microbioreactors. (a) As dimensions of devices scale down to microscopic scales, surfaces and volumes
scale at different rate, so that smaller
devices have a much greater surface/
volume ratio. This change affects
many physical phenomena such as
capillary force (c) and produces
effects such as laminar flow (d),
which allows two liquids to flow
side by side with minimal turbulent
mixing. These microscale phenomena can be useful for microbioreactor opera tion. For example,
capillary force can be used to pump
liquid through microchannels, and
laminar flows can be used to
selectively deliver different reagents to different parts of single
cells. (b) Microfabrication can
further increase the surface/volume
ratio, in a biomimetic manner, by
creating microporous structures.
This is useful since many biological
reactions occur on surfaces.
into smaller ones. Directly downsizing systems and components, however, compromise their functions. Operations
at microscales are often altered in unexpected ways compared to those at macroscales because the relative importance of physical effects that determine the functions of the
microbioreactor components is size dependent. For example, if the linear dimensions of an object are reduced
equally by a factor of 100, the surface area decreases by
a factor of 10,000, and the volume decreases by a factor of
1,000,000 (Fig. 3a). The resulting increase in surface/
volume ratio leads to prominence of surface effects such
as surface tension and viscous drag over gravity and
momentum. Thus, at the microscale, laminar flow, diffusive mixing, and capillary forces dominate over turbulence,
convective mixing, and gravitational forces (Fig. 3c, d) (6).
Scaling laws affect almost every aspect of microbioreactor
operation. For example, it is more difficult to perform
turbulent mixing at the microscale compared to at the
macroscale (79), but more convenient to use electroosmosis flow to transport fluid samples through microchannels
(10). More details of how scaling laws affect the design and
operation of the microbioreactor components are noted in
each section below.
Microbioreactor Operation Principles
The main challenge in bioreactor operation lies in its
dynamic nature. Optimal reaction conditions must be
maintained even as multiple parameters, such as
concentration of molecules, activity of enzymes, quantity
and quality of cells, and heat production are changing.
Understanding the reaction mechanisms is helpful for
optimizing reactor designs. It is also desirable to constantly monitor and control reaction environments in
real time.
Mass And Heat Balance. In bioreactor processes, it is
common to assume steady state, that is, a balance of input,
output flow of materials, and thermal homeostasis. As
substrate input is converted into product, mass balances
need to be closely monitored and controlled. In many
enzymatic processes, for example, the reactions are reversible or are inhibited by products. In such cases, it is
beneficial to favor the forward enzymatic reaction by keeping the upstream reagent concentrations high and constantly removing downstream products. This type of
balance can be achieved to different extents by fed-batch
systems or continuous flow systems that constantly replenish and remove reactor contents. Even in so-called batch
reactors, where there is no flow of liquid in and out of the
reaction chamber, it is often crucial to aerate and maintain
sufficient levels of oxygen at all times.
As dimensions are diminished, the conductive resistance at the channel wall/fluid interface decreases, and
temperature gradient increases, leading to a greater efficiency of thermal conductivity (11). Because heat flux
scales with surface area while heat capacity scales with
volume, thermal time constants and heat flux scale linearly
with decreasing dimensions. The small thermal mass and
associated fast thermal response allows for quick temperature control. On the other hand, since the microsystems
can be heated and cooled quickly, it is difficult to maintain a
constant temperature; external temperatures must be set
at the desired temperature, or constant heating and cooling
MICROBIOREACTORS
of the microbioreactor will be required to avoid fluctuations.

A phenomenon that may not impact macroscopic bioreactors considerably, but has a significant effect on microbioreactors, is evaporation. Unless carefully monitored,
evaporation will greatly affect reactant concentrations
and osmolarity because of the small volume that a microbioreactor can hold.
A variety of mathematical models have been developed
to analyze and control bioreactor thermodynamics and
stoichiometry. Interested readers are referred to existing
texts on the topic (12). For example, Roels (13,14) developed correlations to estimate some of the important thermodynamic parameters to predict yields, substrate and
oxygen requirements, and heat dissipation for cellular
reactions.
Reaction Kinetics. Bioreaction kinetics depends critically on temperature, pH, dissolved oxygen concentrations,
and presence of inhibitors or enhancers. Oxygen fluctuation affects the metabolic and signaling pathways of cells.
For bacterial bioreactors, the main concern is often to
achieve a high enough oxygen transfer rate to match the
oxygen uptake rate and maintain sufficiently high dissolved oxygen concentrations. For mammalian cell cultures, it is necessary to maintain an appropriate
dissolved oxygen concentration. In expansion of stem cells,
for example, lower oxygen concentrations that mimic physiological values gives higher yields and purity of cells
(15,16). Microbioreactors present unique challenges and
opportunities in terms of oxygen transfer. Microbial and
eukaryotic cell cultures require constant replenishment of
dissolved oxygen into the medium because of the low
solubility of oxygen in aqueous solutions (8 mg/L1 at
35 8C in distilled water) (17). For larger bioreactors, bubbling oxygen or passing ambient gas through the cell
culture suspension are commonly used (18). Microscale
gas formation is more challenging, but has been demonstrated elegantly using electrolysis by Maharbiz et al. (19)
(see Components section for details). For microbioreactors
with submilliliter volumes, bubbles are generally avoided
because they can clog channels and are difficult to dislodge
from microchannel walls. For such systems, it is common to
use a gas-permeable membrane to oxygenate the culture
medium (20,21). The large surface/volume ratio of
microdevices provides an advantage for such membranebased oxygen transfer processes. Poly(dimethylsiloxane)
(PDMS), a biocompatible polymer material commonly used
for microdevice fabrication, is particularly advantageous in
this regards due to its high permeability to oxygen.
Enzyme performance and cell growth and function are
particularly sensitive to pH changes. When the reaction is
not within the optimal pH range, the reaction rate declines
dramatically (12). Substrates, products, or contaminants
can reversibly or irreversibly affect enzyme activity. Surface interactions can also alter the kinetics of enzyme
reactions, stability of enzymes (22), as well as growth
and function of cells (23). Precisely dispensing nano- or
picoliters of acids and bases into the tiny reaction chamber
to adjust pH is technically challenging. Furthermore,
because it is difficult to perform convective mixing in
387
microbioreactors, there may be local variations in the pH

and in the reaction kinetics.
Mechanical Considerations. When microbioreactors are
coupled with microfluidic systems, cells are subjected to
shear stresses. For a given maximum flow velocity, smaller
channels will give rise to larger shear stresses (1).
Although excess forces can damage cells, moderate shear
stresses on genetically engineered cells have actually
enhanced the production yield of recombinant proteins
(24). Cells in the body are also often exposed to stretching
and/or compression. These forces are critical factors that
regulate function of cells in muscles, hearts, blood vessels,
lungs, and other tissues (25,26). For example, shear stress
regulates morphology, gene expression and function of
endothelial cells, the monolayers of cells lining the inside
of blood vessels (2730). Shear stress is physiologically
important for maintaining vascular homeostasis (31), regenerate bone and healing fractures (32), differentiate
embryonic stem cells into cardiovascular fate, leukocytes
rolling and tethering to endothelial cells (33). Nuclear
factor-kb (NF-kb), for example, has been identified as a
shear stress- responsive transcription factor that enhances
the transcription of many genes including cytokines,
growth factors, adhesion molecules, and immunoreceptors
in response to shear stress (34).
Some of the biochemical and structural responses of
cells to stretch include enhanced expression of endothelin
(35), nitric oxide (36), and integrin (37) in vascular
endothelial cells, and increased extracellular matrix production in cardiac fibroblasts (38) and in smooth muscle
cells (39). Mechanical loading influence cellular functions
in vitro, including proliferation (40,41), differentiation
(42), hypertrophy (40,43), alignment (41,44), G protein
activation, second messenger activity (45), and gene
expression (43,46).
Cells in the body are constantly subjected to physical
forces, such as cyclic mechanical deformation involving
tension, compression, shear stress, or all three. Cell proliferation, differentiation, migration, signal transduction,
and gene expression are all affected by such mechanical
forces (47). Although it is largely unknown how mechanical
stimuli are converted into intracellular signals of gene
expression (48), there are many efforts to recreate physiological mechanical signals in vitro, either in the form of
shear stresses from fluid flow, hydrodynamic pressures, or
mechanical stresses applied to cells through the substrates. An ideal microbioreactor would provide optimal
biomechanical and biodynamic stimuli for cell and tissue
growth.
Material Considerations. The material used to fabricate
the reaction chambers is crucial because they directly
contact cells, enzymes, and products of bioreactors. Sometimes, it is also necessary to mimic physiological cellular
environments (Fig. 2d) by altering the properties of the
chamber surface to resemble that of the extracellular
matrix (ECM). Proteins or enzymes can be immobilized
onto surfaces to manipulate cell growth. Topographical
features can also be incorporated to further mimic the
ECM environment.
388
MICROBIOREACTORS
Surface Chemistry. As the size of devices decreases, their

surface/volume ratio increases, making surface properties
increasingly important in defining the performances of
smaller bioreactors because cell function is intimately
linked to the properties of the surfaces to which cells
attach. Depending on the application, it is necessary to
promote specific adsorption of proteins that mediate cell
attachment and growth onto surfaces, or to prevent adsorption of proteins and cells. Because cellular receptors that
bind to surfaces are nanoscale in size, it is also important to
be able to pattern adhesive surfaces with resolutions from
microns to nanometers (2,4951). Surface properties are
also important for enzyme-based microbioreactors because
surface properties alter enzyme activity and stability.
Enzymes are commonly immobilized through physical
adsorption or covalent binding onto high surface area
materials, such as carbon, silica, and polymers. Use of
immobilized enzymes is often favored over free enzymes
because of reduction in enzyme costs, ease of recovery,
stability, and ability to be incorporated into microsystems.
Ratners recent review covers important topics including
surface modification of materials to prevent nonspecific
protein adsorption, immobilizing functional groups on surface, and development of synthetic materials (52).
A useful model surface for studying biomaterials interactions is a self-assembled monolayer (SAM). Often formed
on gold using alkane thiols, SAMS are highly ordered
arrays of linear molecules that have one end attached to
the surface of gold or other bulk materials and the other
end exposed to the environment. A useful feature of SAMs
is that their surface properties are determined predominantly by the very terminal functional group. By altering
the nature of these terminal groups, SAMs can prevent
protein and cell attachment or promote binding of specific
ones (53). Other types of systems that are useful for controlling protein adsorption include covalent bonding, via
silanes, to silica or metal. Micropatterns of biopolymers can
also be generated using photolithography, which uses patterns of light to induce region-selective chemical reactions
on a surface. A more recent technique is microcontact
printing, which involves transfer of small molecules or
proteins onto solid substrates from an elastomeric stamp
(53). Microfluidic networks have also been used for protein
patterning: elastomers with embedded channel features
are used to direct small volumes of protein solutions into
networks of channels to create protein patterns corresponding to the path of fluid flow (5456). Surface patterned microfluidic channels can then be used directly as
microbioreactors in which micropatterned cell culture can
be performed.
Topography Control. Living organisms are not flat. The
endothelial lining of blood vessels, for example, exhibit an
irregular wave-like topography to prevent build-up of fatty
deposits (57). Endothelial monolayers have been modeled
as a wavy surface by computational methods to estimate
the influence of the waviness on local flow forces (58).
Muscle fibers form microscale ridges and grooves onto
which myoblasts can attach and proliferate (59). Microfabricated topographical features, regardless of whether
they mimic physiological microtopographies or not, can be
used in microbioreactors to modulate adhesion, align or

orient cells, and even affect cell growth and differentiation.
Examples of topographical features that have been studied
include single cliffs, grooves/ridges, spikes, hills, tunnels
and tubes, fibers, cylinders, mesh, waves, and random
roughness. Materials used for topographical controls
include gold, silicon, carbon, inorganic compounds, such
as silica, lithium niobate, silicon nitride, and polymers,
such as polymethylmethacrylate, silicones, cellulose acetate, collagen, fibrin, and PDMS (51). Microtopography can
also affect surface wetting and fluidic flow patterns. Even
for a simple groove structure, variations of the aspect ratio
and contact angle of the underlying substrate materials
can dramatically alter the morphology of liquid droplets
contacting the surface (60).
Sterilization. Similar to larger scale bioreactors, microbioreactors and their solutions and gases can be sterilized
using heat, chemicals, radiation, or through the filtration
of agents. The small reaction volumes of microbioreactors
provide an advantage in terms of sterilization, because for
a given concentration, the total number of cells, spores, or
other contaminants depends on the volume. Then, assuming that the death rate of the contaminant is independent
of reactor size or contaminant numbers, the sterilization
time needed to extinct the contaminants will decrease with
decreasing reactor size. In this respect, microbioreactors
are more time efficient in batch sterilization than macroscopic ones. Ultraviolet (UV) sterilization is particularly
convenient for transparent microbioreactors, such as those
fabricated in PDMS. When pH sensors or dissolved oxygen
sensors are packaged into the microbioreactor, autoclave
and UV radiation may not be feasible, and alternative
methods may need to be used. Flowing 70% ethanol
through the chambers/channels and subsequently drying
is also sufficient for many microbioreactor applications.
Other Considerations. Phototrophic microorganisms
consume light energy to survive; they can grow in simple
and inexpensive nutrient media. The light requirements
of phototrophic microorganisms, however, impose other
significant constraints on photobioreactor designs (61).
These constraints are due to an exponential attenuation
of the light flow passing through an optically absorbing
medium. Microbioreactors, given their large surface/
volume ratios, are promising for photobioreactor applications. Cultivation of phototrophic microorganisms in optimized photobioreactors would increase the product yield
several folds by maintaining the culture under appropriate conditions.
Ultrasound irradiation can change both the structure
and function of biologic molecules such as proteins and
deoxyribonucleic acid (DNA) (62). At mild intensity, ultrasound irradiation can increase the activity of free enzymes
(63). For example, significant enhancement of reaction rate
can be achieved when exposing subtilisin power to ultrasound irradiation (64). Low energy ultrasound wave irradiation can also optimize the efficiency in ethanol
production by yeast from mixed office waste paper in
bioreactors (65). A variety of microscale ultrasound
systems have been reported. Advanced microbioreactor
MICROBIOREACTORS
systems with ultrasound components may be developed to

make bioreactors more efficient.
MICROBIOREACTOR COMPONENTS
Components For Heat Transfer
External heating elements can be incorporated into microbioreactors to control temperatures. External controls of
temperature include the use of standard cell culture incubators that can accommodate the whole microbioreactor
(66), heating tapes, and water-to-water heat exchangers
(67). Internal, embedded heaters can be comprised of
materials such as thin platinum films (68) or optically
transparent indiumtin oxide (ITO) (69). Temperature
measurements can also be made on chip using metallic,
semiconducting, or optical materials.
An important application where heaters are crucial and
microbioreactors have an advantage is the polymerase
chain reaction (PCR), a temperature-controlled and
enzyme-mediated DNA amplification technology (70). Polymerase chain reaction requires multiple cycles of high, low,
and medium temperatures to separate DNA strands,
anneal the primer to the template DNA, and make complementary copies of the template DNA. Since microsystems usually have high heat conductivity and low heat
capacity, the time for raising and lowering the temperature
is shortened and time will be saved when cyclic temperature fluctuations during the PCR reaction is necessary.
Northrup et al. (71) integrated microfabricated polysilicon
heaters into a micromachined silicon reaction chamber.
Schneegass et al. (68) used a thermocycler chip with integrated thin platinum film heaters and sensors for temperature. Kopp et al. (72) used external copper blocks and
heating cartridges with the surface temperature monitored
by a platinum thin-film resistor. Burns and co-workers (73)
developed a very simple and elegant PCR device that
utilizes RayleighBenard convectiona steady, buoyancy-driven circulatory flow that occurs between two surfaces, one on top and one on the bottom, maintained at two
fixed temperaturesto perform temperature cycling.
Components For Aeration
Electrolytic gas generation provides a compact, scalable
approach for gas delivery to microbioreactors (19). In brief,
a pair of interdigitated Ti/Pt electrodes hydrolyzes electrolyte to generate oxygen gases at the narrow end of a
gradually widened hydrophilic microchannel. The oxygen
bubbles move along the conical microchannels and transfer
into culture medium because of the positive pressure built
up during gas generation and the different surface tension
forces at the front and back of the bubbles formed in the
gradually widening channel. The rate of oxygen generation
can be precisely controlled by pulse width modulation of
the electrode potential.
The smaller the bioreactor, the more crucial it is to avoid
bubble formation to prevent blockage of microchannels.
For such systems, it is advantageous to use gas permeable
membranes that allow diffusion of gases through it without
introduction of bubbles. Because the surface/volume ratio
389
is large in microsystems, oxygen transfer through gaspermeable membranes is often sufficient to ensure adequate oxygenation for biomass production. A straightforward method to fabricate gas-permeable membranes
is to spincoat PDMS prepolymer onto silanized silicon
wafers, cure and harden to generate a thin membrane,
then peel it off (20,21).
Components For Fluid Control
Valves. Valves can be categorized into active and
passive valves. A passive valve is a flow-dependent obstruction that functions without any external actuation. Passive
valves are mostly unidirectional. In an active valve, fluid
flow is directed by active actuation (74). The advantage
that active valves have over passive valves is the degree of
control one has over the timing, rate, and direction of fluid
flow. This type of control is necessary to make real-time
adjustments and for feedback control of microbioreactors.
Although a large variety of valves have been reported, it is
still a challenge to integrate multiple valves into a functional bioreactor system. Difficulties arise because many
valves are incompatible with other components to be integrated, or because the valves require large external systems for actuation.
Passive valves have been used for restricting flow to
one direction, removing air from liquid, or making flows
stop at select channel regions. Although the level of
control is lower compared to active valves, passive valves
have the advantages of having few or no moving parts, less
complexity, easy fabrication, and less chance to break due
to fatigue (75). Recent approaches for passive control
valves involve the use of hydrophobic materials, surface
patterning, and changing channel fluid resistance (by
changing channel geometry) (11). Passively moving micropiston valves have also been fabricated in situ inside
microchannels using laser polymerization of a nonstick
polymer (76).
Most conventional active microvalves couple a flexible
diaphragm (77,78) to, thermopneumatic (79), piezoelectric
(80), electrostatic, electromagnetic (81), bimetallic, or
other types of actuators. The scaling of these actuation
forces to the microscale, however, is often unfavorable and
requires macroscale external actuators for operation. An
interesting alternative to active valves is the use of autonomously regulated valves made of hydrogels that swell in
response to pH or other specific chemical or thermal
environment (11,82). The volume changes of the hydrogel
can valve or obstruct fluid flow directly, or indirectly,
through deformation of a thin PDMS membrane. The
PDMS, when deformed, partially occludes an orifice to
regulate the feedback stream of compensating buffer solution (17). By altering their chemistry, hydrogels can also
valve in response to the changes of temperature, light,
electric fields, carbohydrates, or antigens. Ehrick et al.
(83) incorporated genetically engineered proteins within
hydrogels that swell in response to various ligands as
potential valves for microfluidic channels. Other types
of valves include the use of commercially available Braille
displays (84) or a pneumatic valve system to deform flexible microchannels (85).
390
MICROBIOREACTORS
Pumps. A micropump should ideally be able to pump a

wide range of fluids and gases, be self-priming, and be programmable. Ideal micropumps are still lacking; thus, many
microfluidic applications use macroscopic pumps, such as
syringe pumps. Micropumps can be classified into two main
types: mechanical pumps and nonmechanical pumps (71).
Mechanical pumps use electromagnetic, piezoelectric,
pneumatic, shape memory, electrostatic, thermopneumatic, or thermomechanic components to deliver fluid.
Mechanical pumps provide higher control over average
flow rates, but the flow is often pulsed and the fabrication
relatively complex (53,85). Many types of nonmechanical
micropumps have also been successfully developed. The
flow from nonmechanical pumps is usually pulse-free with
a wide range of flow rates at low pressures and the fabrication is often less complex compared to mechanical pumps.
An electrokinetic pump that utilizes an electric field for
pumping conductive fluids by electrophoresis or electroosmotic flow (EOF) is the most common method to control
flow in microfluidic systems (71). Electrokinetic pumps
have the advantages of direct control, fast response, and
simplicity. However, the substrate material, joule heating
effect, and microchannel charge have to be considered.
Other types of nonmechanical pumps include electrohydrodynamic pumps that use electrostatic forces acting on
dielectric fluids, phase-transfer pumps that use pressure
gradient between gas and liquid phases, electrowetting
fluid actuation systems that use interfacial forces, electrochemical pumps that use the pressure of gas bubbles
generated by electrolysis of water (71), magnetohydrodynamic pumps (86), capillary force, gravity-driven pumps
(87), pneumatic pumps (85), and pumps that use action
of piezoelectric pin arrays in refreshable Braille displays
(84).
Multiple Laminar Streams. For most flows in small channels, viscous forces dominate; thus flow is laminar and
lacks turbulence. When two or more streams pass through
microchannels, they flow in parallel as if they are separated by physical boundaries. Laminar flow is a challenge
for mixing, but a useful phenomenon for microscale fluid
patterning (Fig. 3d) (55,88). By taking advantage of diffusive mixing (but not turbulent mixing) and laminar flows,
spatiotemporally defined gradients can be generated and
have been used to study chemotaxis (89,90). Multiple
laminar flows have also been used for developing microfluidic assay systems [i.e., T-sensor (91)], and studying
subcellular processes when the interfaces of the laminar
streams are positioned over a single cell (92).
Mixers. Mixing is challenging in microfluidic systems
because laminar flows preclude turbulent mixing (Fig. 3d).
Microscale mixers, therefore, generally use elongational
flows or laminar shears to increase interfacial areas
between different fluids and mixing by diffusion. Distributive mixing physically splits the fluid streams into smaller
segments and redistributes them to reduce the striation
thickness. Passive and active mixers have been developed
for microfluidic systems, including laminating mixers (93),
rotary mixers (94), mixing based on out of phase forward
and backward pumping of different liquids (84), plume
mixers (nozzle arrays), chaotic advection mixers (9,95),

movement of liquid plugs (96), and an electroosmotically
driven micromixer that uses multiple intersecting channels to enhance lateral transport (97). Thorough reviews on
micromixers have been given by Hessel et al. (98) and
Nguyen and Wu (99).
Components For Mechanical Stimulation
Shear. Methods commonly used to impart shear stress
on cells include cone viscometers, parallel plates, and
capillary tube flow chambers (100,101). Gradients of shear
stress can also be generated in a curved D-shape microchannel (102). Flow-induced cytoskeleton rearrangements
were shown to depend on the geometry of the channel
(D-shape channel versus flat surfaces, representing experience in microcirculation and large veins, respectively) and
the presence of inflammatory drugs (103).
Stretch. When subjecting cultured cells to mechanical
stretch, proper design and application of a strain device are
required to provide a well-defined and reproducible strain
field to study mechanotransduction. Information from such
in vitro models, which facilitate systematic variations in
mechanical conditions and allow rapid analyses, would
yield tremendous insights into mechanical parameters
that may be important in vivo (104). Biaxial cell strain
devices have been used to strain lung cells (105,106) by
repeated mechanical deformations of a membrane on
which cells are attached. Strain could also be applied using
a magnetic force (107), or via uniaxial cyclical stretch (108).
Components For Separation And Purification
Lysing. Cell contents are separated from their surrounding environment by a cell membrane. The membrane
and an underlying cytoskeletal network provide mechanical strength to the cell and preserve its integrity. The first
step in many analyses or isolation of cell contents is to
disrupt the cell membrane. There are a variety of mechanical (homogenization, milling, ultrasonic disruption, and
blenders) and nonmechanical (detergent, organic solvent,
osmotic shock, enzymatic permeabilization, electrical
discharge, heating, and pressure cycling) methods for disrupting cell membranes on the macroscopic scale. Demonstration of cell lysis on the microscopic scale inside
microfluidic devices, however, has mostly been with nonmechanical methods that use detergents (96,109), electrical discharges, or lasers (110). Miniaturized cell
electrolysis devices can work with small amounts of cells
and reduce the amount of purification compared to other
protocols (111,112). For example, Waters et al. (113) developed a microchip that is capable of performing Escherichia
coli lysis, PCR amplification, and eletrophoretic analysis
on a single device.
Separation. Cell separation techniques are fundamental to clinical diagnosis, therapy, and biotechnological
production (114). For example, it is crucial to have
purified cells before proliferation and production of
cellular products. Current approaches include optical
tweezers (115), centrifugation (116,117), filtration
MICROBIOREACTORS
(109,116,118), fluorescence-based cell sorting (FACS)

(119,120) or magnetically activated cell sorting (MACS)
(121), electric field-based manipulations and separations
(114,122124), and cell-motility based sorting (125,126).
Microtechnology opens new opportunities in cell and biomolecule sorting that take advantage of laminar flow
behaviors (125), electrical field properties (127), or other
microscale phenomena. It also provides the opportunity to
combine multiple modes of separation into an integrated
system. Researchers have used physiological fluid mechanical phenomenon observed in blood microcirculation
(plasma skimming and leukocyte margination) to filter
leukocytes from whole blood (128). Petersson et al. (129)
combined acoustic wave forces and laminar flow to continuously sort erythrocytes from lipid particles in the whole
blood. Because these two components were different in
density and responsiveness to pressure, erythrocytes were
enriched at the pressure node (along the center of the
channel) and lipid particles were gathered at the pressure
antinodes (along the side walls). Finally, the erythrocytes
and lipid microemboli separated into different branches at
the end of the main channel. Because of the variety of
different properties by which cells of interest need to be
sorted, it is important to develop multiple modes of cell
sorting. Reviews about microfluidic cell analysis and sorting devices can be found elsewhere (130,131).
Filtration. Microfabrication techniques have been
developed to integrate filters and membranes inside microbioreactors. Zhao et al. (132) and Hisamoto et al. (133)
successfully produced semipermeable nylon membranes
inside microfluidic channels, by taking advantage of laminar flow so that a polymerization reaction would occur at
the liquid interface of the two flows and produce a thin
membrane in predetermined areas of a microfluidic device.
Components For Monitoring And Control
A key requirement for microbioreactors is the ability to
measure parameters, such as temperature, dissolved oxygen, pH, and flow rate. For systems involving cells, mass
balances are even more complex than with enzyme bioreactors because of the larger number of products and
byproducts produced and the complex responses of cells
to the changes in material concentrations. In small
volumes, it is difficult or impossible to use standard,
macroscopic, industrial probes. Miniaturized sensors must
be developed (17). Sensors that do not consume the analytes are also preferred to avoid depletion and also because
sample extraction is hard to achieve in closed and compact
microsystems without disrupting the devices.
Optical, electrochemical, and thin-film solid-state conductivity are the three main categories of microsensing
schemes. Many of the most useful microdetection schemes
are based on optical measurements such as fluorescence
intensity (134), fluorescence lifetime, chemiluminescence
(135), and bioluminescence (136). Optical sensing is convenient because molecular or nanoscale sensors are simple to introduce inside microchannels and readout can be
detected from a distance without direct external contacts.
In addition, many optical probes can sense without con-
391
suming oxygen or perturbing pH. Nonperturbing sensors

are important for maintaining a constant microenvironment because the quantities of chemicals are small in
microbioreactors. Bioluminescence and chemiluminescence are sensitive with the detection limits down to
10181021 mol, which offers great advantage over other
spectroscopic-based detection mechanisms. Laser-induced
fluorescence detections in microsystems have been
reviewed by Johnson and Landers (137). Other optical
detection methods for microfluidic systems have been
reviewed by Mogensen et al. (138).
Solgel-based probes encapsulated by biologically localized embedding (PEBBLEs) allow real-time measurement
of molecular oxygen, pH, and ions inside and around living
cells (139). Kostov et al. (17) used an optical sensing system
integrated with semiconductor light sources and detectors
to perform continuous measurements of pH, optical density, and dissolved oxygen in miniature bioreactors. A
drawback of optical sensing is the need for light sources,
lenses for focusing, and detectors, which are more challenging to miniaturize compared to the sensor probes themselves. A useful solution is to use optical fiber-based
systems, which allows decoupling of the probes from the
light sources and detectors, and enables detection at sites
inaccessible by conventional spectroscopic sensors (140).
Compared with silicon micromachining techniques, the
fabrication and integration of polymeric optical elements
(waveguides, lenses and fiber-to-waveguide couplers) with
microfluidic channels are fast and simple (141). An oxygensensitive fluorescent dye has been developed to monitor
dissolved oxygen levels in a system. This provides advantages over electrochemically based sensors [for a review,
see Ref. 142] due to their size, ease of fabrication, and
sensitivity (141).
Electronic microsensor is another category that contains a large number of useful biochemical detectors
(143). Electronic microsensors usually require direct hard
wiring of sensors to a readout system but can be easier to
multiplex and are often smaller overall compared to optical
systems. For example, Walther et al. (144) integrated a pHISFET (ion-sensitive field-effect transistor), a temperature-sensitive diode, and a thin-film platinum redox electrode on a single chip. The chip is mounted on a carrier and
inserted into the chamber to monitor various biological
parameters such as pH and redox potential. Brown and coworkers developed polymer membrane-based solid-state
sensors to measure pH, and ions (145).
Microfabricated ultrasensitive nanocalorimeters can
measure heat generation to monitor cell metabolic activity
in response to agonist and antagonist using as few as 10
cells and without prior knowledge of the mode of action of
these drugs. This measurement is noninvasive and quantitative and is envisioned to be useful for pharmaceutical
companies to find drug candidate (146).
Li et al. (147) developed a microfabricated acoustic wave
sensor to measure the stiffness of a single cell. This sensor
is promising for drug screening and toxicology studies. For
example, the acoustic wave sensor is envisioned to measure
the rigidity of a heart cell to distinguish the effect of
positive and negative ionotropic drugs. Positive ionotropic
drugs are useful for treating congestive heart failure and
392
MICROBIOREACTORS
negative ionotrpic drugs for hypertension. The acoustic

wave sensor is also interesting for single cell muscle physiology.
Cells themselves can be used to sense and amplify
biological signals. Several groups have developed histamine sensors by integrating cells on microfluidic devices
(148,149). This chip-based detector caters to the need for a
simple, rapid, and safe method for allergy identification.
Cells can be engineered to have a variety of biological
recognition events coupled to reporter genes to specifically
sense analytes of interest (150).
Some new exciting prospects for future biosensors are
in the area of nanotechnology. For example, quantum dots
(151) and nanoscale PEBBLES (139) are extending the
limits of sensitivity, stability, biocompatibility, and flexibility in optical sensing. Quantum dots are small semiconductor nanocrystals (on the order of nanometers to a
few micrometers). Fluorescent quantum dots are able to
detect biological species by fluorescing only when coming
in contact with viable cells, making them useful probes for
many types of labeling studies. These quantum dots are
photobleached very slowly and can be manufactured to
emit a wide range of wavelengths. They can be used in cell
biology for the labeling of cellular structures, tracking the
fate of individual cells, or as contrast agents (152). Nanowire-based sensors with their small size and higher sensitivity would also be ideal for integration into
microbioreactors (153). Many microsensors have been
developed and are ready for integration into microbioreactors.
Fabrication
A variety of methods and types of substrates are available
for microfabrication. The most traditional and widely used
method of microfabrication is photolithography. Originally
developed for the microelectronics industry, photolithography is precise, highly reproducible, and capable of mass
production. Photolithography uses patterns of UV light
coming through a photomask to area-selectively induce
chemical reactions in a polymeric, light-sensitive photoresist coated onto a semiconductor substrate. During development, the light exposed regions of the photoresist are
selectively removed or selectively left behind generating
micropatterned photoresist structure. The exposed areas of
the substrate are then chemically etched to provide features of various depths and shapes.
Due to the high equipment costs involved in photolithography and because silicon and glass substrates used in
electronic and mechanical devices are not necessarily the
best materials for biological applications, alternative types
of microfabrication have been developed. A cost-effective
and experimentally straightforward method called soft
lithography has been particularly useful for biological
applications (2,53). Soft lithography uses elastomeric
materials, such as PDMS to create microstructures, and
to pattern and manipulate surfaces. The process involves
casting PDMS against a photolithographically defined
master mold to yield a polymeric replica. The PDMS replica
is then sealed against another material to form channels
and reservoirs. Alternatively, the replica can be used in a
technique called microcontact printing, where the PDMS

mold is used as a stamp to transfer protein or molecular ink
to a substrate. The PDMS has several properties, which
make it useful for biological applications: (1) biocompatibility allows cell culture on and inside PDMS structures,
(2) optical transparency allows optical inspection and sensing, (3) gas permeability allows long-term growth of cells
without depletion of oxygen, (4) flexibility allows cell cultured on PDMS to be mechanically stretched (2,50,53).
Other polymer based microfabrication techniques
include hot embossing, injection molding, and laser ablation. A hot embossing technique called nano-imprint lithography developed by Chou et al. (154) has the ability to
fabricate sub-10 nm nanometer features. This process
creates nanostructures in a resist by deforming the resist
shape with embossing (155). Moriguchi et al. (156) developed a unique photothermal microfabrication technique,
where agar microchamber arrays with living cells inside
them can be remolded in situ during cell cultivation.
Integration
The development of a microbioreactor requires assembling
multiple functional units (for electronic, mechanical, biological, and chemical processing) into a compact device.
Integration and packaging poses a whole new challenge on
top of the challenges to develop individual components.
With macroscopic bioreactors, it is relatively straightforward to connect different components together with little
or no worries of space organization. As one scales down,
the placement of components must be performed strategically as the room around the reactor chamber decreases
dramatically (because volume scales as length cubed, a 10
time reduction in linear dimensions, e.g., will lead to a
1000 time reduction in the available space). There are also
technical challenges to fabricate microscale fittings and
connectors, or to join two components via connectors even
if they could be fabricated. A variety of processes used to
build integrated circuits and microelectromechanical systems have played a major role in fabricating integrated
microfluidic systems and microbioreactors. These
technologies, known collectively as micromachinging,
selectively etch away or deposit structural layers on silicone wafers.
Recent efforts to reduce costs, enhance material biocompatibility, and needs for diverse chemical and mechanical
properties have also led to the use of a wide variety of
polymeric materials in microfabricated devices. Integration,
unless planned carefully, can lead to material incompatibilities in fabrication or operation. Notable accomplishments of integrated microfluidic systems include
DNA analysis chips (78,157), pneumatically driven microfluidic cell sorters and protein recrystallization chips
(77,119,158,159), portable cell-based biosensor systems
(160), microfermentors with integrated sensors (17,20,21),
and a computerized microfluidic cell culture system actuated using the pins of a refreshable Braille display (84).
Integrating fluid control components for cell-culture microbioreactors is rapidly progressing, but still underdeveloped. Interested readers are referred to recent review
articles on integrated microfluidic devices (161).
MICROBIOREACTORS
SPECIFIC EXAMPLES OF MICROBIOREACTORS

This section presents select examples of microbioreactors.
The examples are not exhaustive, but are meant to show
representative examples in each of the following four categories: (1) microbioreactors that are used to optimize bioproduction, (2) microbioreactors that provide cells with
physiological microenvironments to more accurately predict physiological drug kinetics and toxicity, (3) microbioreactors that are used to develop cell-based therapies, and
(4) microbioreactors for mechanistic studies. Because the
field is still young, the devices are relatively simple and
many are still prototypes rather than refined products
ready for real world applications. The rapid advances,
however, promise an increasing role of microbioreactors
in the clinic, laboratory, and at home or other points of
need.
Microbioreactors For Optimizing Production Conditions
With the development of microtechnologies, more and
more bioprocess optimization is performed in small
volume bioreactors with integrated detection systems.
For example, Rao and co-workers (17) have demonstrated
parallel fermentations of E. coli. in a milliliter-size microbioreactor. A smaller, microliter-size fermentor has been
developed recently by the Jensens group (20). The performances of these microbioreactors are comparable to traditional liter-size fermentors: Measurements of pH,
dissolved oxygen (DO) and optical density (OD) of biomass
in these microbioreactors have similar profiles as those in
benchtop fermentors. Similarities in cellular metabolism
and growth show the potential of using microbioreactors
for bioprocess optimization. Arrays of microfermentors
with integrated sensors and actuators are envisioned to
drastically reduce the cost and time for developing new
bioprocesses.
Microbioreactors For Toxicological And Drug Testing
Drugs need to be tested for toxicity and efficacy
before administration to humans. Accurate prediction,
however, is a challenge because most current drug tests
are performed only on human cells or animals. Animal
tests are expensive and time consuming, and efficacy of a
drug obtained from an animal surrogate study can still be
difficult to extrapolate to humans (162). Even when one
uses human cells for analysis, the result may be
totally different from what occurs physiologically because
cells cultured in flasks or dishes experience a totally
different microenvironment. Therefore, a microscale
human surrogate with microcirculatory systems, threedimensional (3D) tissue organizations, and appropriate
cellcell and tissuetissue interactions would be beneficial
in predicting human responses to drug treatment
more precisely. Below are two notable examples of such
efforts.
Bioartificial Livers. There are two major applications for
which artificial livers are developed: one is to replace organ
functions in patients with liver failure, and the other is to
perform toxicology testing of drug candidates. Organ repla-
393
cement functions require large bioartificial livers (BALs),

whereas the toxicology studies would benefit from microscale BALs capable of conducting high throughput analyses. Microscale BALs are promising as convenient and
low cost in vitro models for screening drug toxicities particularly in light of the fact that approximately one-half of all
drug toxicities involve the liver.
Liver failure is the seventh leading cause of death by
disease in the United States. About 26, 000 people died
each year because of liver failure. Transplantation is limited by the supply of donor organs and the cost of the
surgery (163). Extracorporeal BAL devices have been proposed as substitutes for transplantation. Macroscopic
BALs have been tested in clinical trials (164). In an attempt
to maximize the efficacy of the BALs, and to develop in vitro
liver systems for biological and toxicological studies, several groups have microfabricated liver cell culture systems
(165).
Microbioreactors for liver cell cultures have several
configurations, ranging from flat-plate (163) or matrixsandwiched monolayer designs (166) to 3D perfusion cultures (165). A flat-plate microbioreactor with an oxygen
permeable membrane was shown to support viability and
synthetic functions of hepatocytes cocultured with 3T3-J2
fibroblasts (163). This microchannel bioreactor was also
functional when connected extracorporeally to a rat (162).
The results indicate that this device can potentially be
used as a liver support device and for the eventual scale-up
to clinical devices. Compared with other configurations,
monolayer designs excel in mass transfer, easy fabrication, and easy optical analysis of cells (165), although cells
might be damaged by exposure to shear stress (167).
Griffith and co-workers (165) developed an array of microbioreactors (with each channel 300 300 230 mm,
L W H in dimension) that support 3D culture of liver
cells by perfusion. Liver cells cultured in this device
showed viable tissue structures and tight junctions, glycogen storage, and bile canaliculi. Membrane-based 3D perfusion hepatocyte culture systems have also been
developed (66).
Micro CCA. It is important to integrate cells from
different tissues together to simulate physiological
drug metabolism. Shuler and co-workers developed Cell
Culture Analogs (CCAs) that combine mathematical
pharmacokinetics models with cell culture-based experimental studies to mimic human responses. The CCAs
have compartmentalized cell cultures representing different tissues. Interconnections between these compartments allow circulation of media and metabolites. Since
the CCAs mimic the time-dependent exposure and the
metabolic interaction between multiple types of tissues
and cells, predictions from the CCAs may be more accurate compared to existing in vitro models. Shuler and coworkers proved the concept of CCA with a macroscopic
three-compartment (liver, lung, and other tissues) system
and showed the feasibility and potential usefulness of
such devices in testing naphthalene toxicity (168,169).
MicroCCAs with three (liver, lung, and other tissues) and
four (liver, lung, fat, and other tissues) compartments
have also been reported (170,171).
394
MICROBIOREACTORS
Microbioreactors For Therapeutical Applications

Microfluidic Systems As Assisted Reproductive Technologies. Microdevices provide unique platforms for artificial
reproduction and may ultimately increase the efficiency,
safety, and cost-effectiveness of in vitro fertilization
procedures. Currently, many embryo experiments are performed in macroscopic culture dishes, where human or
animal oocytes (eggs) and embryos are manipulated manually. Use of pipettes for cellular manipulations is labor
intensive and low in accuracy, reproducibility, and efficiency. In addition, the practice of transferring embryos
from one type of media into another is abrupt and may
shock the embryo due to the sudden change of environment
(172174); inside the female tract, the supply of nutrients,
growth factors, and hormones changes gradually as the
embryo development progresses. An alternative to manual
pipetting is the use of microfluidic channels. Microfluidics
is ideal for use in artificial reproduction, because it is a
procedure that occurs physiologically inside small tubes
and ducts, the size and numbers of cells (oocytes, sperms)
required match well with dimensions of microsystems.
Beebe et al. (175) demonstrated manipulations of
embryos and oocytes within microfluidic channels. The
microfluidic systems can transport single mouse embryos
through a channel network (176), remove the zona pellucida by chemical treatment (177), remove cumulus cells
from oocytes via mechanical suction (178), and culture
embryos in static or dynamic fluid environments (179).
In some cases, embryos cultured in microfluidic channels
develop faster compared to embryo grown in culture dishes
and with growth kinetics that are closer to what is observed
in vivo.
Cho et al. (125) developed a Microscale Integrated
Sperm Sorter (MISS) that isolates motile sperms based
on the ability of the motile sperms, but not the nonmotile
ones, to cross-laminar flow streamlines. The device allows
small volume samples that are difficult to handle with
conventional sperm-sorting techniques to be sorted efficiently using a mild biomimetic sorting mechanism. The
MISS integrates power source, sample injection ports, and
sorting channel into one disposable polymer device making
the device potentially useful not only clinically, but also as
an at-home male infertility test (180).
Implantable Microcapsules. Microbioreactors that produce and release natural or recombinant bioagents are
useful for delivering therapeutic agents in vivo to enhance
metabolic function (181) or treat neurological disorders
(182) and cancers (183,184). Mammalian cells, plant cells,
microorganisms, enzymes, and biochemical compounds
have been encapsulated in a variety of semipermeable
containers mainly made of synthetic and natural hydrogels
(e.g., poly(vinyl alchohol), poly(hydroxyl ethyl methacrylate), calcium alginate k-carrageenan, chitosan, collagen,
and gelatin). Here we focus on the application of microencapsulated mammalian cells. The outer membranes of
the microspheres encapsulating the cells serve as selective
barriers that allow exchange of nutrients, wastes, and
therapeutic agents but block the passage of encapsulated
cells as well as macromolecular components of the immune
system. This approach has been effective in delivering

genetically engineered cells that secrete growth hormone
to partially correct growth retardation (185), recombinant
human bone morphogenetic protein-2 (rhBMP-2) to induce
bone formation and regeneration (186), interleukin-2 to
delay tumor progression and prolong survival (187), coagulation factor IX for treatment of hemophilia B (188),
dopamine to treat Parkinsons disease (182), insulin to
maintain blood glucose level (189), and analgesic substances to relief pain (190). Most of these works are aided
by using murine and canine models. Some of the most
advanced systems are in early clinical trials. Biocompatibility (191,192) mechanical stability of the capsule material (193195), efficacy (196), safety (197), and cost are still
under evaluation and optimization. Reviews of therapeutic
uses of microencapsulated genetically engineered cells can
be found elsewhere (198).
Microbioreactors For Understanding Biological Responses
Use Of Multiple Laminar Streams To Study Subcellular
Biology And Chemotaxis. Physiological cell environments
are heterogeneous with local production and consumption
of key growth factors, nutrients, and signaling molecules.
Microfluidic systems are useful for mimicking such micropatterns of chemicals around cells. A particularly simple
and useful method is to take advantage of small channel
dimensions to generate multiple laminar streams that flow
in parallel and adjacent to each other inside the same
microchannel with minimum mixing (Fig. 3d). Such techniques can be even used to treat different parts of single
living cells with different small molecular drugs, proteins,
and small particles such as low density lipoprotein (LDL)
(88).
Cancer and normal cells receive similar local stimuli
inside the body, but behave totally differently. A critical
question is why these differences arise. Taking advantage
of multiple laminar flows to perform subcellular epidermal
growth factor (EGF) stimulation, Sawano et al. (92)
revealed differences in signal propagation between carcinoma and normal cells in response to local EGF stimulation.
Many cells direct their motion in response to chemical
gradients. This phenomenon, called chemotaxis, protects
microorganisms by allowing them to move toward more
favorable conditions. In mammals, chemotaxis is important for guiding cell migration during development,
embryogenesis, cancer metastasis, and inflammation. A
challenge for analyzing chemotaxis is the lack of methods
to generate well-defined chemical gradients that are stable
and do not change with time. This difficulty arises due to
the diffusivity of molecules and resulting changes of concentration profiles over time. Jeon et al. (90) demonstrated
the use of microfluidic systems with branched networks of
channels to generate stable gradients of interleukin-8
(IL-8) with linear, parabolic, and periodic concentration
profiles. The position and shape of the concentration gradients were controlled by adjusting the flow rates and the
positions to which reagent of interest were added into the
channel network (89,199). The well-defined gradients
allowed straightforward quantification of chemotaxis
MICROBIOREACTORS
coefficients as well as to observe complex migration behaviors of leukocytes in response to different concentration
gradient profiles.
Studies Of Vascular Diseases In Microfluidic Channels.
The mechanical forces that accompany blood flow and
pressure fluctuation influence vascular cellular biology
and pathology in many ways. As the blood flow along a
vessel, the viscous drag forces constantly expose endothelial cells (ECs) to shear stress. The pulsatility of the blood
flow also induces a periodic change of circumferential
strain on ECs and their underlying smooth muscle cells
(SMCs). These mechanical forces have been found to cause
important biological changes in endothelial cell morphology and function, such as alignment and elongation
(30,200), low density lipoprotein uptake (201), tissue plasminogen activator synthesis and secretion (202), and proliferation (67). There has also been studies about the
combined effect of shear stress and cyclic strain on ECs
(203205) and SMCs (206208). While much has been
revealed about the alterations in EC function induced by
mechanical stresses, relatively little is known about the
mechanism mechanical signaling.
Capillary-size microfluidic channels were used to model
malaria, a potentially vital disease caused by loss of
deformability of red blood cells due to P. falciparum parasites infection. Erythrocytes exhibited increased rigidity
and decreased deformability with the progression of the
disease, as demonstrated by their increased difficulties to
transverse through 2 to 8 mm wide PDMS channels. This
type of microfluidic system may potentially be useful to
screen antimalaria drugs (209).
CONCLUSION AND FUTURE PROSPECTS
The convergence of bioreactors with advances in microtechnology is starting an exciting revolution in medicine.
With microfabrication technologies, many copies of a
device can be generated and operated with quick procedures and reduced cost. The ability to perform parallel
assays allows high throughput optimization of bioprocess
conditions, opening the way for cost-efficient biopharmaceuticals production.
Humans and other living organisms are inherently
microscopic in their essence, being comprised of networks
of microscopic reactors (i.e., the cells), and interconnected
by microfluidic vasculatures. Efforts to miniaturize bioreactors would lead to not only the development of smaller
pharmaceutical production and screening systems, but
also the construction of more physiological in vitro cell
culture systems where the ultimate goal is to develop
microbioreactors as animal or human surrogates. Even
for cullture of single cell types, the ability of microfluidic
systems to simulate physiological microenvironments is
useful for revealing disease mechanisms, drug testing,
use as biosensors, and single-cell-based clinical procedures
such as in vitro fertilization. More complex microbioreactor
systems with multiple cell types are being developed for
toxicology studies. So-called animals-on-a-chip or minihumans provides exciting prospects for efficient drug discovery and personalized medicine.
395
Current state-of-the-art microbioreactors are still relatively simple with few components and limited sensing and
control. Many are highly specialized and nonroutine in
their use. The overall footprints of the systems are also
often still macroscopic. Current advances in micro- and
nanotechnologies as well as in medicine, however, promise
rapid improvements in performance, accessibility, and
sophistication of microbioreactors. Current trends point
to a future where the gap between manmade devices
and living organisms will narrow and applications of microbioreactors to medicine will grow.
ACKNOWLEDGMENTS
We thank the Whitaker Foundation, National Science
Foundation (BES-0238625, DMI-0403603), National Institutes of Health (EB00379-01, HD049607-01), NASA
(NNC04AA21A), and the U.S. Army Research Laboratory
and the U.S. Army Research Office under contract/grant
number DAAD19-03-1-0168 for financial support. We
thank G. D. Smith, S. C. Chang, H. Chen, P.Y. Choi,
B. H. Chueh, T. Kable, E. O. Kass, K. Ke, J.H. Kim,
S. M. K. Lau, E.C. Lee, W.H. Lee, S.R. Mandal, J.A. Miller,
Y. Murgha, S.O. Charoen, A.B. Ozel, S.J. Segvich, P.D.
Settimi, D. Sud, B. Teply, M. Zhang, C. Zhong for
assistance.
BIBLIOGRAPHY
1. Walker GM, Zeringue HC, Beebe DJ. Microenvironment
design considerations for cellular scale studies. Lab Chip
2004;4:9197.
2. Shim J, et al. Micro- and nanotechnologies for studying cellular function. Curr Top Med Chem 2003;3:687703.
3. Bhadriraju K, Chen CS. Engineering cellular microenvironments to cell-based drug testing improve. Drug Discov Today
2002;7:612620.
4. Desai TA, et al. Microfabricated immunoisolating biocapsules.
Biotechnol Bioeng 1998;57:118120.
5. Krenkova J, Foret F. Immobilized microfluidic enzymatic
reactors. Electrophoresis 2004;25:35503563.
6. Madou MJ. Fundamentals of Microfabrication. Boca Raton
(FL): CRC Press; 2002.
7. Tabeling P, et al. Chaotic mixing in cross-channel micromixers. Philos Trans R Soc London Ser A-Math Phys Eng Sci
2004;362:9871000.
8. Stremler MA, Haselton FR, Aref H. Designing for chaos: Applications of chaotic advection at the microscale. Philos. Trans R
Soc Lond Ser A-Math Phys Eng Sci 2004;362:10191036.
9. Stroock AD, et al. Chaotic mixer for microchannels. Science
2002;295:647651.
10. Squires TM, Bazant MZ. Induced-charge electro-osmosis. J
Fluid Mech 2004;509:217252.
11. Ehrfeld W, Hessel V, Lowe H. Microreactors: New Technology
for Modern Chemistry. Chichester : Wiley-VCH; 2000.
12. McDuffie NG. Bioreactor Design Fundamentals. Boston:
Butterworth-Heinemann; 1991.
13. Roels JA. Macroscopic Thermodynamics and the Description of
Growth and Product Formation in Microorganisms. Washington, (D.C.): American Chemical Society; 1983.
14. Roels JA. Energetics and Kinetics in Biotechnology. Amsterdam: Elsevier Biomedical Press; 1983.
396
MICROBIOREACTORS
15. Ezashi T, Das P, Roberts RM. Low O-2 tensions and the
prevention of differentiation of hES cells. Proc Natl Acad Sci
U S A 2005;102:47834788.
16. Csete M, et al. Oxygen-mediated regulation of skeletal muscle satellite cell proliferation and adipogenesis in culture. J
Cell Physiol 2001;189:189196.
17. Kostov Y, Harms P, Randers-Eichhorn L, Rao G. Low-cost
microbioreactor for high-throughput bioprocessing. Biotechnol Bioeng 2001;72:346352.
18. Blanch HW, Clark DS. Biochemical Engineering. New York:
Marcel Dekker; 1997.
19. Maharbiz MM, et al. A microfabricated electrochemical oxygen generator for high-density cell culture arrays. J Microelectromech Syst 2003;12:590599.
20. Zanzotto A, et al. Membrane-aerated microbioreactor for
high-throughput bioprocessing. Biotechnol Bioeng 2004;
87:243254.
21. Maharbiz MM, Holtz WJ, Howe RT, Keasling JD. Microbioreactor arrays with parametric control for high-throughput experimentation. Biotechnol Bioeng 2004;85:376
381.
22. Irazogui G, Villarino A, Batista-Viera F, Brena BM. Generating favorable nano-environments for thermal and solvent
stabilization of immobilized beta-galactosidase. Biotechnol
Bioeng 2002;77:430434.
23. Hara M, Adachi S, Higuchi A. Enhanced production of carcinoembryonic antigen by CW-2 cells cultured on polymeric
membranes immobilized with extracellular matrix proteins.
J Biomater Sci-Polym Ed 2003;14:139155.
24. Keane JT, Ryan D, Gray PP. Effect of shear stress on expression of a recombinant protein by Chinese hamster ovary cells.
Biotechnol Bioeng 2003;81:211220.
25. Dardik A, et al. Shear stress-stimulated endothelial cells
induce smooth muscle cell chemotaxis via platelet-derived
growth factor-BB and interleukin-1 alpha. J Vasc Surg
2005;41:321331.
26. Kher N, Marsh JD. Pathobiology of atherosclerosisa
brief review. Semin Thromb Hemostasis 2004;30:665
672.
27. Butcher JT, Penrod AM, Garcia AJ, Nerem RM. Unique
morphology and focal adhesion development of valvular
endothelial cells in static and fluid flow environments. Arterioscler Thromb Vasc Biol 2004;24:14291434.
28. Kladakis SM, Nerem RM. Endothelial cell monolayer formation: Effect of substrate and fluid shear stress. Endothelium
2004;11:2944.
29. Imberti B, Seliktar D, Nerem RM, Remuzzi A. The response
of endothelial cells to fluid shear stress using a co-culture
model of the arterial wall. Endothelium-New York 2002;9:
1123.
30. Song J, et al. A Computer-Controlled Microcirculatory Support System for Endothelial Cell Culture and Shearing. Anal
Chem 2005, In press.
31. Krizanac-Bengez L, Mayberg MR, Janigro D. The cerebral
vasculature as a therapeutic target for neurological disorders
and the role of shear stress in vascular homeostatis and
pathophysiology. Neurol Res 2004;26:846853.
32. Richards M, et al. Bone regeneration and fracture healing
Experience with distraction osteogenesis model. Clin Orthop
Related Res 1998; S191S204.
33. Kim MB, Sarelius IH. Role of shear forces and adhesion
molecule distribution on P-selectin-mediated leukocyte rolling in postcapillary venules. Amer J Physiol-Heart Circ Phy
2004;287:H2705H2711.
34. Blackwell TS, Christman JW. The role of nuclear factorkappa B in cytokine gene regulation. Amer J Respir Cell
Molec Biol 1997;17:39.
35. Awolesi MA, Sessa WC, Sumpio BE. Cyclic Strain up-Regulates Nitric-Oxide Synthase in Cultured Bovine Aortic
Endothelial-Cells. J Clin Invest 1995;96:14491454.
36. Suzuki N, et al. Up-regulation of integrin beta (3) expression
by cyclic stretch in human umbilical endothelial cells. Biochem Biophys Res Commun 1997;239:372376.
37. Booz GW, Baker KM. Molecular signaling mechanisms controlling growth and function of cardiac fibroblasts. Cardiovasc Res 1995;30:537543.
38. Kolpakov V, et al. Effect of mechanical forces on growth and
matrix protein-synthesis in the in-vitro pulmonary-artery
analysis of the role of individual cell-types. Circ Res
1995;77:823831.
39. Desrosiers EA, Methot S, Yahia LH, Rivard CH. Response of
ligamentous fibroblasts to mechanical stimulation. Ann Chir
1995;49:768774.
40. Komuro I, et al. Mechanical loading stimulates cell hypertrophy and specific gene-expression in cultured rat cardiac
myocytespossible role of protein-kinase-C activation. J Biol
Chem 1991;266:12651268.
41. Neidlingerwilke C, Wilke HJ, Claes L. Cyclic stretching of
human osteoblasts affects proliferation and metabolisma
new experimentalmethod and its application. J Orthopaed
Res 1994;12:7078.
42. Vandenburgh HH, Karlisch P. Longitudinal growth of
skeletal myotubes in vitro in a new horizontal mechanical
cell stimulator. In Vitro Cell Develop Biol 1989;25:607
616.
43. Reusch P, et al. Mechanical strain increases smooth muscle
and decreases nonmuscle myosin expression in rat vascular
smooth muscle cells. Circ Res 1996;79:10461053.
44. Stauber WT, Miller GR, Grimmett JG, Knack KK. Adaptation of rat soleus muscles to 4-wk of intermittent strain.
J Appl Physiol 1994;77:5862.
45. Gudi SRP, Lee AA, Clark CB, Frangos JA. Equibiaxial
strain and strain rate stimulate early activation of G proteins
in cardiac fibroblasts. Amer J Physiol-Cell Physiol 1998;
43:C1424C1428.
46. Vandenburgh HH, Hatfaludy S, Sohar I, Shansky J. Stretchinduced prostaglandins and protein-turnover in cultured
skeletal-muscle. Amer J Physiol 1990;259:C232C240.
47. Cowan DB, Lye SJ, Langille BL. Regulation of vascular
connexin43 gene expression by mechanical loads. Circ Res
1998;82:786793.
48. Davies PF, Flow-mediated endothelial mechanotransduction. Physiol Rev 1995;75:519560.
49. Blawas AS, Reichert WM. Protein patterning. Biomaterials
1998;19:595609.
50. Zhu XY, et al. Fabrication of reconfigurable protein matrices
by cracking. Nat Mater 2005.
51. Zhu XY, Bersano-Begey TF, Takayama S. Nanomaterials for
Cell Engineering. In: Nalwa HS. editor, Encyclopedia of
Nanoscience and Nanotechnology. American Scientific Publishers; 2004. p 857-878.
52. Ratner BD, Bryant SJ. Biomaterials: Where we have been
and where we are going. Annu Rev Biomed Eng 2004;6:41
75.
53. Whitesides GM, et al. Soft lithography in biology and biochemistry. Annu Rev Biomed Eng 2001;3:335373.
54. Delamarche E, et al. Microfluidic networks for chemical
patterning of substrate: Design and application to bioassays.
J Am Chem Soc 1998;120:500508.
55. Takayama S, et al. Patterning cells and their environments
using multiple laminar fluid flows in capillary networks. Proc
Natl Acad Sci U S A 1999;96:55455548.
56. Folch A, Toner M. Cellular micropatterns on biocompatible
materials. Biotechnol Prog 1998;14:388392.
MICROBIOREACTORS
57. Grigioni M, et al. Pulsatile flow and atherogenesis: Results
from in vivo studies. Int J Artif Organs 2001;24:784792.
58. Satcher RL, Bussolari SR, Gimbrone MA, Dewey CF. The
distribution of fluid forces on model arterial endothelium
using computational fluid-dynamics. J Biomech Eng
1992;114:309316.
59. Evans DJR, Britland S, Wigmore PM. Differential response
of fetal and neonatal myoblasts to topographical guidance
cues in vitro. Dev Genes Evol 1999;209:438442.
60. Seemann R, et al. Wetting morphologies at microstructured
surfaces. Proc Natl Acad Sci USA 2005;102:18481852.
61. Ogbonna JC, Tanaka H. Night biomass loss and changes in
biochemical composition of cells during light/dark cyclic culture of Chlorella pyrenoidosa. J Ferment Bioeng 1996;82:
558564.
62. Macleod RM, Dunn F. Ultrasonic irradiation of enzyme solutions. J Acoust Soc Amer 1966;40:1202.
63. Ishimori Y, Karube I, Suzuki S. Acceleration of immobilized
alpha-chymotrypsin activity with ultrasonic irradiation.
J Mol Catal 1981;12:253259.
64. Vulfson EN, Sarney DB, Law BA. Enhancement of subtilisincatalyzed interesterification in organic-solvents by ultrasound irradiation. Enzyme Microb Technol 1991;13:123126.
65. Wood BE, Aldrich HC, Ingram LO. Ultrasound stimulates
ethanol production during the simultaneous saccharification
and fermentation of mixed waste office paper. Biotechnol
Prog 1997;13:232237.
66. Ostrovidov S, Jiang JL, Sakai Y, Fujii T. Membrane-based
PDMS microbioreactor for perfused 3D primary rat hepatocyte cultures. Biomed Microdevices 2004;6:279287.
67. Geiger RV, Berk BC, Alexander RW, Nerem RM.
Flow-induced calcium transients in single endothelialcellsspatial and temporal analysis. Amer J Physiol
1992;262:C1411C1417.
68. Schneegass I, Brautigam R, Kohler JM. Miniaturized flowthrough PCR with different template types in a silicon chip
thermocycler. Lab Chip 2001;1:4249.
69. Shivashankar GV, Liu S, Libchaber A. Control of the expression of anchored genes using micron scale heater. Appl Phys
Lett 2000;76:36383640.
70. Shen KY, Chen XF, Guo M, Cheng J. A microchip-based PCR
device using flexible printed circuit technology. Sensors and
Actuators B-Chem 2005;105:251258.
71. Northrup MA, Ching MT, White RM, Watson RT. DNA
amplification with a microfabricated reaction chamber. Proc
IEEE Int Conf Solid-state Sensors Actuators 1993; 924926.
72. Kopp MU, de Mello AJ, Manz A. Chemical amplification:
Continuous-flow PCR on a chip. Science 1998;280:10461048.
73. Krishnan M, Ugaz VM, Burns MA. PCR in a RayleighBenard convection cell. Science 2002;298:793793.
74. Elwenspoek M, Lammerink TSJ, Miyake R, Fluitman JHJ.
Towards integrated microliquid handling systems. J Micromechanic Microengineer 1994;4:227245.
75. Lao AIK, Lee TMH, Hsing IM, Ip NY. Precise temperature
control of microfluidic chamber for gas and liquid phase
reactions. Sensors and Actuators A-Phys 2000;84:1117.
76. Altman GH, et al. Advanced bioreactor with controlled application of multi-dimensional strain for tissue engineering.
J Biomech Eng 2002;124:742749.
77. Thorsen T, Maerkl SJ, Quake SR. Microfluidic large-scale
integration. Science 2002;298:580584.
78. Grover WH, et al. Monolithic membrane valves and diaphragm pumps for practical large-scale integration into glass
microfluidic devices. Sensors and Actuators B-Chem 2003;89:
315323.
79. Vandepol FCM, Wonnink DGJ, Elwenspoek M, Fluitman
JHJ. A thermo-pneumatic actuation principle for a micro-
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
397
miniature pump and other micromechanical devices. Sensors

and Actuators 1989;17:139143.
Smits GJ. A piezoelectric micropump with three valves working peristalitically. Sensors and Actuators 1990;15:203206.
Bosch D, et al. A silicon microvalve with combined electromagnetic/electrostatic actuation. Sensors and Actuators APhys 1993;37-8:684692.
Liu RH, Yu Q, Beebe DJ. Fabrication and characterization of
hydrogel-based microvalves. J Microelectromech Syst 2002;
11:4553.
Ehrick JD, et al. Genetically engineered protein in hydrogels
tailors stimuli-responsive characteristics. Nat Mater 2005;4:
298302.
Gu W, et al. Computerized microfluidic cell culture using
elastomeric channels and Braille displays. Proc Natl Acad Sci
U S A 2004;101:1586115866.
Unger MA, et al. Monolithic microfabricated valves and
pumps by multilayer soft lithography. Science 2000;288:
113116.
Lemoff AV, Lee AP. An AC magnetohydrodynamic micropump. Sens Actuators B-Chem 2000;63:178185.
Zhu XY, et al. Arrays of horizontally-oriented mini-reservoirs
generate steady microfluidic flows for continuous perfusion
cell culture and gradient generation. Analyst 2004;129:1026
1031.
Takayama S, et al. Selective chemical treatment of cellular
microdomains using multiple laminar streams. Chem Biol
2003;10:123130.
Dertinger SKW, Chiu DT, Jeon NL, Whitesides GM. Generation of gradients having complex shapes using microfluidic networks. Anal Chem 2001;73:12401246.
Jeon NL, et al. Neutrophil chemotaxis in linear and complex
gradients of interleukin-8 formed in a microfabricated device.
Nat Biotechnol 2002;20:826830.
Hatch A, et al. A rapid diffusion immunoassay in a T-sensor.
Nat Biotechnol 2001;19:461465.
Sawano A, Takayama S, Matsuda M, Miyawaki A. Lateral
propagation of EGF signaling after local stimulation is
dependent on receptor density. Dev Cell 2002;3:245257.
Bessoth FG, deMello AJ, Manz A. Microstructure for efficient
continuous flow mixing. Anal Commun 1999;36:213215.
Hong JW, et al. A nanoliter-scale nucleic acid processor with
parallel architecture. Nat Biotechnol 2004;22:435439.
Liu RH, et al. Passive mixing in a three-dimensional serpentine microchannel. J Microelectromech Syst 2000;9:190
197.
Song H, Tice JD, Ismagilov RF. A microfluidic system for
controlling reaction networks in time. Angew Chem Int Ed
Engl 2003;42:768772.
Burke BJ, Regnier FE. Stopped-flow enzyme assays on a chip
using a microfabricated mixer. Anal Chem 2003;75:17861791.
Hessel V, Lowe H, Schonfeld F. Micromixersa review on
passive and active mixing principles. Chem Eng Sci
2005;60:24792501.
99. Nguyen NT, Wu ZG. Micromixersa review. J Micromechanic Microengineer 2005;15:R1R16.

100. Frangos JA, McIntire LV, Eskin SG. Shear-Stress Induced
stimulation of mammalian-cell metabolism. Biotechnol
Bioeng 1988;32:10531060.
101. Blackman BR, Barbee KA, Thibault LE. In vitro cell shearing
device to investigate the dynamic response of cells in a controlled hydrodynamic environment. Ann Biomed Eng
2000;28:363372.
102. Frame MDS, Chapman GB, Makino Y, Sarelius IH. Shear
stress gradient over endothelial cells in a curved microchannel system. Biorheology 1998;35:245261.
398
MICROBIOREACTORS
103. Frame MD, Sarelius IH. Flow-induced cytoskeletal changes

in endothelial cells growing on curved surfaces. Microcirculation 2000;7:419427.
104. Chien S, Li S, Shyy JYJ. Effects of mechanical forces on
signal transduction and gene expression in endothelial cells.
Hypertension 1998;31:162169.
105. Clark CB, Burkholder TJ, Frangos JA. Uniaxial strain
system to investigate strain rate regulation in vitro. Rev
Sci Instr 2001;72:24152422.
106. Buck RC. Reorientation response of cells to repeated stretch
and recoil of the substratum. Exp Cell Res 1980;127:470
474.
107. Mourgeon E, et al. Mechanical strain-induced posttranscriptional regulation of fibronectin production in fetal lung
cells. Amer J Physiol-Lung Cell M Ph 1999;277:L142L149.
108. Kato T, et al. Up-regulation of COX2 expression by uni-axial
cyclic stretch in human lung fibroblast cells. Biochem Biophys Res Commun 1998;244:615619.
109. Schilling EA, Kamholz AE, Yager P. Cell lysis and protein
extraction in a microfluidic device with detection by a
fluorogenic enzyme assay. Anal Chem 2002;74:17981804.
110. Soughayer JS, et al. Characterization of cellular optoporation with distance. Anal Chem 2000;72:13421347.
111. Lee SW, Tai YC. A micro cell lysis device. Sens Actuators
A-Phys 1999;73:7479.
112. Heo J, Thomas KJ, Seong GH, Crooks RM. A microfluidic
bioreactor based on hydrogel-entrapped E. coli: Cell viability, lysis, and intracellular enzyme reactions. Anal Chem
2003;75:2226.
113. Waters LC, et al. Microchip device for cell lysis, multiplex
PCR amplification, and electrophoretic sizing. Anal Chem
1998;70:158162.
114. Huang Y, et al. Electric manipulation of bioparticles and
macromolecules on microfabricated electrodes. Anal Chem
2001;73:15491559.
115. Leitz G, Weber G, Seeger S, Greulich KO. The laser
microbeam trap as an optical tool for living cells. Physiol
Chem Phys Med Nmr 1994;26:6988.
116. Hogman CF. Preparation and preservation of red cells. Vox
Sang 1998;74:177187.
117. Bauer J. Advances in cell separation: Recent developments
in counterflow centrifugal elutriation and continuous flow
cell separation. J Chromatogr B 1999;722:5569.
118. Cheng J, Kricka LJ, Sheldon EL, Wilding P. Sample preparation in microstructured devices, microsystem technology in chemistry and life science. Top Curr Chem 1998;
215231.
119. Fu AY, et al. A microfabricated fluorescence-activated cell
sorter. Nat Biotechnol 1999;17:11091111.
120. Rathman M, et al. The development of a FACS-based strategy for the isolation of Shigella flexneri mutants that are
deficient in intercellular spread. Mol Microbiol
2000;35:974990.
121. Handgretinger R, et al. Isolation and transplantation of
autologous peripheral CD34() progenitor cells highly purified by magnetic-activated cell sorting. Bone Marrow
Transplant 1998;21:987993.
122. Volkmuth WD, Austin RH. DNA electrophoresis in microlithographic arrays. Nature(London) 1992;358:600602.
123. Li PCH, Harrison DJ. Transport, manipulation, and reaction of biological cells on-chip using electrokinetic effects.
Anal Chem 1997;69:15641568.
124. Wang XB, et al. Cell separation by dielectrophoretic fieldflow-fractionation. Anal Chem 2000;72:832839.
125. Cho BS, et al. Passively driven integrated microfluidic
system for separation of motile sperm. Anal Chem
2003;75: 16711675.
126. Horsman KM, et al. Separation of sperm and epithelial cells

in a microfabricated device: Potential application to forensic
analysis of sexual assault evidence. Anal Chem 2005;77:
742749.
127. Chou CF, et al. Electrodeless dielectrophoresis of single- and
double-stranded DNA. Biophys J 2002;83:21702179.
128. Shevkoplyas SS, Yoshida T, Munn LL, Bitensky MW. Biomimetic autoseparation of leukocytes from whole blood in a
microfluidic device. Anal Chem 2005;77:933937.
129. Petersson F, et al. Continuous separation of lipid particles
from erythrocytes by means of laminar flow and acoustic
standing wave forces. Lab Chip 2005;5:2022.
130. Minc N, Viovy JL. Microfluidics and biological applications:
the stakes and trends. C R Phys 2004;5:565575.
131. Huh D, et al. Microfluidics for flow cytometric analysis of
cells and particles. Physiol Meas 2005;26:R1R26.
132. Zhao B, Viernes NOL, Moore JS, Beebe DJ. Control and
applications of immiscible liquids in microchannels. J Am
Chem Soc 2002;124:52845285.
133. Hisamoto H, et al. Chemicofunctional membrane for integrated chemical processes on a microchip. Anal Chem 2003;
75:350354.
134. Kawabata T, Washizu M. Dielectrophoretic detection of molecular bindings. IEEE Trans Ind Appl 2001;37:16251633.
135. Wu XZ, Suzuki M, Sawada T, Kitamori T. Chemiluminescence on a microchip. Anal Sci 2000;16:321323.
136. Roda A, et al. Biotechnological applications of bioluminescence and chemiluminescence. Trends Biotech 2004;22:295
303.
137. Johnson ME, Landers JP. Fundamentals and practice for
ultrasensitive laser-induced fluorescence detection in
microanalytical systems. Electrophoresis 2004;25:3513
3527.
138. Mogensen KB, Klank H, Kutter JP. Recent developments in
detection for microfluidic systems. Electrophoresis 2004;25:
34983512.
139. Xu H, et al. A real-time ratiometric method for the determination of molecular oxygen inside living cells using solgel-based spherical optical nanosensors with applications to
rat C6 glioma. Anal Chem 2001;73:41244133.
140. Wolfbeis OS. Fiber-optic chemical sensors and biosensors.
Anal Chem 2002;74:26632677.
141. Chang-Yen DA, Gale BK. An integrated optical oxygen
sensor fabricated using rapid-prototyping techniques. Lab
Chip 2003;3:297301.
142. Vandaveer WR, et al. Recent developments in electrochemical detection for microchip capillary electrophoresis. Electrophoresis 2004;25:35283549.
143. Bakker E, Telting-Diaz M. Electrochemical sensors. Anal
Chem 2002;74:27812800.
144. Walther I, et al. Development of a miniature bioreactor for
continuous-culture in a space laboratory. J Biotechnol
1994;38:2132.
145. Yoon HJ, et al. Solid-state ion sensors with a liquid
junction-free polymer membrane-based reference electrode for blood analysis. Sens Actuators B-Chem 2000;
64:814.
146. Johannessen EA, et al. Micromachined nanocalorimetric
sensor for ultra-low-volume cell-based assays. Anal Chem
2002;74:21902197.
147. Li PCH, Wang WJ, Parameswaran M. An acoustic wave
sensor incorporated with a microfluidic chip for analyzing
muscle cell contraction. Analyst 2003;128:225231.
148. Kurita R, et al. Differential measurement with a microfluidic device for the highly selective continuous measurement
of histamine released from rat basophilic leukemia cells
(RBL-2H3). Lab Chip 2002;2:3438.
MICROBIOREACTORS
149. Matsubara Y, et al. Application of on-chip cell cultures for
the detection of allergic response. Biosens Bioelectron
2004;19: 741747.
150. Daunert S, et al. Genetically engineered whale-cell sensing
systems: Coupling biological recognition with reporter
genes. Chem Rev 2000;100:27052738.
151. Wu XZ, et al. Immunofluorescent labeling of cancer marker
her2 and other cellular targets with semiconductor quantum dots. Nat Biotechnol 2003;21:4146.
152. Parak WJ, Pellegrino T, Plank C. Labelling of cells with
quantum dots. Nanotechnology 2005;16:R9R25.
153. Cui Y, Wei QQ, Park HK, Lieber CM. Nanowire nanosensors
for highly sensitive and selective detection of biological and
chemical species. Science 2001;293:12891292.
154. Chou SY, Krauss PR, Renstrom PJ. Imprint lithography
with 25-nanometer resolution. Science 1996;272:8587.
155. Raiteri R, Grattarola M, Butt HJ, Skladal P. Micromechanical cantilever-based biosensors. Sens Actuators B-Chem
2001;79:115126.
156. Moriguchi H, et al. An agar-microchamber cell-cultivation
system:flexible change of microchamber shapes during cultivation by photo-thermal etching. Lab Chip 2002;2:125130.
157. Burns MA, et al. An integrated nanoliter DNA analysis
device. Science 1998;282:484487.
158. Hansen CL, Skordalakes E, Berger JM, Quake SR. A robust
and scalable microfluidic metering method that allows protein crystal growth by free interface diffusion. Proc Natl
Acad Sci USA 2002;99:1653116536.
159. Chou HP, Spence C, Scherer A, Quake S. A microfabricated
device for sizing and sorting DNA molecules. Proc Natl Acad
Sci USA 1999;96:1113.
160. DeBusschere BD, Kovacs GTA. Portable cell-based biosensor system using integrated CMOS cell-cartridges. Biosens
Bioelectron 2001;16:543556.
161. Erickson D, Li DQ. Integrated microfluidic devices. Anal
Chim Acta 2004;507:1126.
162. Shito M, et al. In vitro and in vivo evaluation of albumin
synthesis rate of porcine hepatocytes in a flat-plate bioreactor. Artif Organs 2001;25:571578.
163. Tilles AW, et al. Effects of oxygenation and flow on the
viability and function of rat hepatocytes cocultured in a
microchannel flat-plate bioreactor. Biotechnol Bioeng
2001;73:379389.
164. Nyberg SL, et al. Primary hepatocytes outperform Hep G2
cells as the source of biotransformation functions in a
bioartificial liver. Ann Surg 1994;220:5967.
165. Powers MJ, et al. Functional behavior of primary rat liver
cells in a three-dimensional perfused microarray bioreactor.
Tissue Eng 2002;8:499513.
166. Bader A, et al. Development of a small-scale bioreactor for
drug metabolism studies maintaining hepatospecific functions. Xenobiotica 1998;28:815825.
167. Allen JW, Bhatia SN. Improving the next generation of
bioartificial liver devices. Semin Cell Dev Biol 2002;
13:447454.
168. Ghanem A, Shuler ML. Combining cell culture analogue
reactor designs and PBPK models to probe mechanisms of
naphthalene toxicity. Biotechnol Prog 2000;16:334345.
169. Ghanem A, Shuler ML. Characterization of a perfusion
reactor utilizing mammalian cells on microcarrier beads.
Biotechnol Prog 2000;16:471479.
170. Park TH, Shuler ML. Integration of cell culture and microfabrication technology. Biotechnol Prog 2003;19:243253.
171. Viravaidya K, Shuler ML. Incorporation of 3T3-L1 cells to
mimic bioaccumulation in a microscale cell culture analog
device for toxicity studies. Biotechnol Prog 2004;20:590
597.
399
172. Kruip TAM, Bevers MM, Kemp B. Environment of oocyte

and embryo determines health of IVP offspring. Theriogenology 2000;53:611618.
173. Bavister BD. Interactions between embryos and the culture
milieu. Theriogenology 2000;53:619626.
174. Fukui Y, Lee ES, Araki N. Effect of medium renewal during
culture in two different culture systems on development to
blastocysts from in vitro produced early bovine embryos. J
Anim Sci 1996;74:27522758.
175. Beebe D, et al. Microfluidic technology for assisted reproduction. Theriogenology 2002;57:125135.
176. Glasgow IK, et al. Handling individual mammalian embryos
using microfluidics, IEEE Trans Biomed Eng 2001;48:570
578.
177. Zeringue HC, Wheeler MB, Beebe DJ. Zona pellucida
removal of mammalian embryos in a microfluidic systems.
Micro Total Analysis Syst 2000; 214217.
178. Zeringue HC, Beebe DJ, Wheeler MB. Removal of cumulus
from mammalian zygotes using microfluidic techniques.
Biomed Microdevices 2001;3:219224.
179. Hickman DL, Beebe DJ, Rodriguez-Zas SL, Wheeler MB.
Comparison of static and dynamic medium environments
for culturing of pre-implantation mouse embryos. Comparative Med 2002;52:122126.
180. Suh RS, et al. Rethinking gamete/embryo isolation and
culture with microfluidics. Hum Reprod Update
2003;9:451461.
181. Korbutt GS, et al. Improved survival of microencapsulated
islets during in vitro culture and enhanced metabolic function following transplantation. Diabetologia 2004;47:1810
1818.
182. Vallbacka JJ, Nobrega JN, Sefton MV. Tissue engineering
as a platform for controlled release of therapeutic agents:
implantation of microencapsulated dopamine producing
cells in the brains of rats. J Control Release 2001;72:93
100.
183. Takenaga M, et al. A single treatment with microcapsules
containing a CXCR4 antagonist suppresses pulmonary
metastasis of murine melanoma. Biochem Biophys Res
Commun 2004;320:226232.
184. Yu BL, Chang TMS. Effects of long-term oral administration of polymeric microcapsules containing tyrosinase on
maintaining decreased systemic tyrosine levels in rats. J
Pharm Sci 2004;93:831837.
185. AlHendy A, Hortelano G, Tannenbaum GS, Chang PL.
Growth retardationan unexpected outcome from growth
hormone gene therapy in normal mice with microencapsulated myoblasts. Hum Gene Ther 1996;7:6170.
186. Zilberman Y, et al. Polymer-encapsulated engineered adult
mesenchymal stem cells secrete exogenously regulated
rhBMP-2, and induce osteogenic and angiogenic tissue formation. Polym Adv Technol 2002;13:863870.
187. Cirone P, Bourgeois JM, Austin RC, Chang PL. A
novel approach to tumor suppression with microencapsulated recombinant cells. Hum Gene Ther 2002;13:1157
1166.
188. Hortelano G, Wang L, Xu N, Ofosu FA. Sustained and
therapeutic delivery of factor IX in nude haemophilia B
mice by encapsulated C2C12 myoblasts: Concurrent
tumourigenesis. Haemophilia 2001;7:207214.
189. Chen JP, et al. Microencapsulation of islets in PEG-amine
modified alginate-poly(L-lysine)-alginate microcapsules for
constructing bioartificial pancreas. J Ferment Bioeng
1998;86:185190.
190. Xue YL, et al. Pain relief by xenograft of subarachnoid
microencapsulated bovine chromaffin cells in cancer
patients. Prog Nat Sci 2000;10:919924.
400
191. Cole DR, et al. Transplantation of microcapsules (a potential

bioartificial organ)biocompatibility and host-reaction. J
Mater Sci-Mater Med 1993;4:437442.
192. Lou WH, Qin XY, Wu ZG. Preliminary research on biocompatibility of alginate-chitosan-polyethyleneglycol microcapsules. Minerva Biotecnol 2000;12:235240.
193. Van Raamsdonk JM, Cornelius RM, Brash JL, Chang PL.
Deterioration of polyamino acid-coated alginate microcapsules in vivo. J Biomater Sci-Polym Ed 2002;13:863
884.
194. Sakai S, Ono T, Ijima H, Kawakami K. Behavior of enclosed
sol- and gel-alginates in vivo. Biochem Eng J 2004;22:19
24.
195. Koch S, et al. Alginate encapsulation of genetically engineered mammalian cells:comparison of production devices,
methods and microcapsule characteristics. J Microencapsul
2003;20:303316.
196. Arica B, et al. Carbidopa/levodopa-loaded biodegradable
microspheres:in vivo evaluation on experimental Parkinsonism in rats. J Control Release 2005;102:689697.
197. Leblond FA, et al. Studies on smaller (similar to 315 (mM)
microcapsules: IV. Feasibility and safety of intrahepatic
implantations of small alginate poly-L-lysine microcapsules. Cell Transplant 1999;8:327337.
198. Chang TMS, Prakash S. Therapeutic uses of microencapsulated genetically engineered cells. Mol Med Today 1998;4:
221227.
199. Jeon NL, et al. Whitesides GM. Generation of solution and
surface gradients using microfluidic systems. Langmuir
2000;16:83118316.
200. Sprague EA, Steinbach BL, Nerem RM, Schwartz CJ. Influence of a laminar steady-state fluid-imposed wall shearstress on the binding, internalization, and degradation of
low-density lipoproteins by cultured arterial endothelium.
201. Diamond SL, Eskin SG, McIntire LV. Fluid-flow stimulates
tissue plasminogen-activator secretion by cultured humanendothelial cells. Science 1989;243:14831485.
202. Levesque MJ, Sprague EA, Schwartz CJ, Nerem RM. The
influence of shear-stress on cultured vascular endothelialcellsthe stress response of an anchorage-dependent mammalian-cell. Biotechnol Prog 1989;5:18.
203. Gomes N, et al. Shear stress modulates tumour cell adhesion to the endothelium. Biorheology 2003;40:4145.
204. Davies PF, Tripathi SC. Mechanical-stress mechanisms and
the cellan endothelial paradigm. Circ Res 1993;72:239
245.
205. Ikeda M, et al. Extracellular signal-regulated kinases 1
and 2 activation in endothelial cells exposed to cyclic
strain. Am J Physiol-Heart Circul Physiol 1999;276:
H614H622.
206. Smith PG, Roy C, Zhang YN, Chauduri S. Mechanical stress
increases RhoA activation in airway smooth muscle cells.
Am J Respir Cell Mol Bio 2003;28:436442.
207. Lee T, Kim SJ, Sumpio BE. Role of PP2A in the regulation
of p38-MAPK activation in bovine aortic endothelial
cells exposed to cyclic strain. J Cell Physiol 2003;194:349
355.
208. Han O, Takei T, Basson M, Sumpio BE. Translocation of
PKC isoforms in bovine aortic smooth muscle cells exposed
to strain. J Cell Biochem 2001;80:367372.
209. Shelby JP, et al. A microfluidic model for single-cell capillary obstruction by Plasmodium falciparum infected erythrocytes. Proc Natl Acad Sci USA 2003;100:1461814622.
See also MICROARRAYS;
NEERING.
MICROFLUIDICS; NANOPARTICLES; TISSUE ENGI-
JULIE A. STENKEN
Rensselaer Polytechnic Institute
Troy, New York
MICRODIALYSIS SAMPLING: NON-SPECIALIST VIEW

Microdialysis sampling devices are minimally invasive
miniature dialyzers that can be implanted into a distinct
tissue region to obtain a chemical snapshot over an integrated time period. In combination with appropriate chemical detection methods for the targeted substances, a
microdialysis sampling device may be considered to be a
universal biosensor. Obtaining chemical information from
different tissues can often lead to either a greater understanding of the underlying chemistry involved with the
physiological function of the organ or the origin of a particular disease process. A simplified view of the microdialysis
sampling device is shown in Fig. 1. The central part of the
microdialysis sampling device is a single semipermeable
hollow fiber membrane with dimensions that range
between 200 and 500 mm for its external diameter and 1
and 30 mm in length. A perfusion solution is passed
through the device at microliter per minute flow rates.
Compounds diffuse from the tissue space into the dialysis
probe and are carried to an outlet to undergo chemical
analysis. Originally microdialysis sampling was developed
to obtain real-time chemical information from rodent brain
and was termed intracranial dialysis. Microdialysis sampling has now been applied for chemical collection from
nearly every single organ. In addition to neurotransmitter
collection, the device has been used for endocrinology,
immunology, metabolism, and pharmacokinetic applications as shown in Table 1. The biomedical literature cites
A AP
A
P
AP
AP
P
P
P
AP
AP
A
P
o.d. 200-500
microns
Length 130 mm
PERFUSATE
~ 0.5 to 2 l/min
A
A
A
A
DIALYSATE
[A]outlet
Blood
Vessel
A
Figure 1. Microdialysis sampling process. A perfusion fluid that
closely matches the ionic strength and composition of the fluid
external to the microdialysis membrane is passed through at flow
rates between 0.5 and 2.0 mLmin1. Analytes, A, that are not
protein bound, AP, diffuse through the extracellular fluid space
(wavy lines) and can pass through the pores of the semipermeable
membrane are collected into the device. The analyte outlet
concentration [A]outlet can then be quantified using a suitable
detection method.
Table 1. Typical Microdialysis Sampling Uses
Collection of endogenous analytes from brain including neurotransmitters (dopamine, norepinephrine, serotonin, glutamate,
GABAa , glucose, peptides and proteins (cytokines).
Collection of endogenous small hydrophilic analytes from other
tissues (e.g., glucose for diabetics).
Collection of peptides and proteins from perphiral tissues, for
example, glutathione, neuropeptides, and different cytokines
and growth factors (e.g., VEGF).
Collection of xenobiotic analytes for pharmacokinetic or pharmacodynamic studies from numerous tissue sites (brain, dermis,
muscle, and tumors) in both animals and humans.
Localized delivery of analytes followed by concomitant recovery of
endogenous analytes or metabolized products (e.g., spin traps,
metabolites).
a
GABA r-aminobutyric acid
thousands of research articles that have used microdialysis

sampling devices to study many different basic and clinical
research problems with perhaps 90% or greater of these
applications focused in neuroscience. As more life scientists
realize that microdialysis sampling devices exist, this
device will be used more often to solve many additional
clinical problems outside of the neurosciences.
INTRODUCTION
Mammalian organs are highly complex systems that
achieve their functions through a multifaceted chemical
communication network. For laboratory studies focused on
understanding these chemical networks, the organ is often
broken down into its component parts (cells, subcellular
components, extracellular matrix components, etc.) to create more controlled conditions. These types of studies have
allowed for a great understanding of the individual component parts, but do not address how the chemical communication may occur within the intact organ. Common
diagnostic collection methods, such as blood or urine sampling, are too far removed from organs to be able to provide
desired information about localized organ biochemistry.
Gaining chemical information from an organ system prior
to the creation of microdialysis sampling devices required
either organ dissection or noninvasive analysis methods.
Removing organs to access chemical content is fraught
with many concerns including the preparation of the sample that involves in most cases sacrifice of the animal (or a
biopsy for humans) as well as concerns about chemical
stability and loss during the sample preparation process.
For nearly all organs, the changes in localized tissue
chemistry caused during sacrifice may severely alter
some chemical communication systems. Finally, organ
dissection does not allow for temporal studies of targeted
analytes.
An increasing number of medical devices and instruments are becoming available to noninvasively measure in
vivo chemical composition at spatially defined sites. Noninvasive measurements typically use spectroscopic instruments that are highly analyte specific. In particular, the
most well-known medical devices for achieving these tasks
401
are positron emission tomography (PET) and magnetic

resonance imaging (MRI). Positron emission tomography
scanning requires production of special isotopes (11C, 13N,
15
O, 18F) with limited half-lives. Similarly, MRI uses 1H to
create an image and is a useful imaging technique because
the concentration of hydrogen nuclei in water and fat is
roughly 100 M (mol/L1). Magnetic resonance spectroscopy (MRS) can be used to noninvasively detect other
isotopes (1H, 13C, 15N, 19F, 31P), but requires very high
concentrations of these nuclei for detection. Fluorescence
imaging has also been used for noninvasive measurements
in conjunction with near-infrared (IR) tags or with enzyme
substrates that become fluorescent when cleaved (13).
While these spectroscopic techniques hold significant promise for some areas of clinical medicine (cardiovascular
and oncology), they are quite limited with respect to the
range of analytes that can be detected and thus potential
applications. While not noninvasive, minimally invasive
microdialysis sampling devices coupled with appropriate
dialysate analytical detection methods allow for measurements of localized tissue chemistry with significantly
greater analyte flexibility and spatial resolution.
Microdialysis sampling originated as an alternative to
pushpull perfusion devices for use in mammalian brain.
Pushpull perfusion devices were used to collect fluid
relevant to synaptic transmission (4). During a pushpull
perfusion, a solution that closely matches the ionic chemical composition of the extracellular fluid (ECF) is loaded
into a syringe and this is gently infused into the implantation site. This perfusion fluid mixes with the existing ECF
and then is pulled back into the device. The analysis of
pushpull samples became a useful tool for tracking neurotransmitter activities coupled with allowing for a
remarkable understanding of the underlying chemical
events associated with a wide variety of behavioral and
physiological stimuli. Insertion of pushpull cannula could
potentially cause tissue damage or lesions, which raised
many concerns among researchers since this type of
damage could limit the usefulness of the neurochemical
data collected. In other words, researchers were often
concerned that sampling was really taking place in a lake
of fluid that may not contain the representative chemical
components of the extracellular fluid space.
The pushpull sampling method was principally applied
to neurochemical sampling. For other tissues, other extracellular fluid sampling methods have been described.
These other methods include open-flow microperfusion
and wick methods. Both of these methods have been used
in muscle, as well as dermal sampling. Additionally, blister
methods are common for obtaining interstitial fluid from
the skin. Open-flow microperfusion is similar to pushpull
perfusion. A cannula device contains an inner tube and an
outer tube with open pores (typically 500 mm) is placed
over this inner tube. A peristaltic pump then simultaneously delivers and withdraws fluid through the device
(5). Open-flow microperfusion has primarily been used for
sampling the extracellular fluid in muscle and skin. Alternatives to perfusion methods are wick methods, which use
nylon wicks to sample the extracellular fluid space in
animals and humans (68). To obtain multiple samples
over a specified time period would require insertion and
402
removal of individual wick devices. This repeated insertion

and removal might cause additional trauma to the sampling site.
To overcome the tissue damage concerns associated
with pushpull perfusion for neuroscience applications,
the use of semipermeable dialysis membranes at the tips
of the pushpull cannula were used (9,10). These original
dialysis bags eventually led to flow-through microdialysis
probes introduced by Pycock and Ungerstedt in 1974 (11).
The advantage of microdialysis sampling over the push
pull cannula is that fluid is not pushed into sensitive brain
tissue. Unlike a pushpull perfusion, microdialysis sampling is a continuous process that provides a sample
that excludes many of the components from the ECF. This
exclusion process serves to provide a relatively clean
sample for chemical analysis. Today, this technique has
been widely used by life scientists to attain site-specific
access to numerous tissue sites to study and solve many
problems, which include, but are not limited to the following applications: (1) To elucidate the role of different
neurotransmitters and neuropeptides in specifically
defined brain regions; (2) To collect glucose continuously
over many days to give a better chemical picture to diabetes
specialists to determine the efficacy of an insulin regimen
for individual patients; (3) To collect energy metabolites
(glucose, lactate, pyruvate) as well as administered drugs
(morphine) to understand diseased energy metabolism and
blood-brain barrier transport from head trauma patients;
(4) To determine if an efficacious drug concentration is
reaching a specific infection site or diseased tissue sites for
antineoplastic therapy or antimicrobial therapy; (5) To
determine pharmacokinetic parameters in single animals;
(6) To determine metabolite formation and accumulation in
various tissues after a drug dose in a single animal over
time; (7) To determine the extent of drug bloodbrain
barrier permeation of new drugs in animal models; (8)
To collect various endogenous peptides and proteins
(e.g., cytokines and growth factors) from different peripheral sites in animals and humans (12).
To newcomers to this device, the principles of microdialysis sampling operation at first seem deceptively simple. At the most basic level, the microdialysis sampling
device may be considered to behave as an artificially
implanted blood vessel that allows free analyte diffusion
into the inner fiber lumen. However, as will be discussed in
this article, numerous considerations that involve both an
understanding of the localized biology and physiology, as
well as the underlying mass transport processes are essential to microdialysis sampling data interpretation.
MICRODIALYSIS SAMPLING PRINCIPLES OF OPERATION

In principle, as long as the microdialysis probe can be
implanted, it can be used for sampling localized tissue
biochemistry. Microdialysis sampling requires only a few
pieces of equipment. This equipment is not cost-prohibitive, which is the reason that many different researchers
can perform microdialysis sampling experiments in their
own laboratories. For most experiments, the necessary
equipment includes a perfusion pump to deliver the perfu-
sion fluid and a microdialysis probe. In addition to the

pump and probe, for animal studies, a device to hold either
an anesthetized animal (e.g., stereotaxic unit for neuroscience procedures) or a bowl with appropriate swivels
to prevent tangled tubing for freely moving animals may be
necessary.
Microdialysis Sampling Instrumentation Components
The basic components needed to perform microdialysis
sampling experiments in an awake-freely moving animals
has been described (13). The components required for this
type of experiment includes the microperfusion pump, an
inlet and outlet fluid swivel that prevents the fluid lines
from becoming tangled, and a bowl system to allow the
animal to freely move. Additional components can include
refrigerated fraction collectors to allow collection and storage of sensitive samples. Microperfusion pumps used to
deliver the perfusion fluid through the microdialysis probe
are capable of delivering volumetric flow rates between 0.1
and 20 mLmin1. Flow rates between 0.5 and 2.0 mLmin1
are commonly used during most microdialysis sampling
experiments.
Microdialysis sampling perfusion fluids are chosen to
closely match the ionic strength and composition of the
external tissue extracellular fluid surrounding the microdialysis probe. Perfusion fluids passed through microdialysis sampling probes are a form Ringers solution for
which there are numerous published chemical compositions (14,15). Typical Ringers solutions contain 150 mM
NaCl, 4 mM KCl, and 2.4 mM CaCl2 and can also be
supplemented with glucose and other ionic salts (MgCl2).
These solutions are used both to maintain fluid balance, as
well as ion balance across the dialysis membrane. Maintaining fluid balance across the dialysis membrane is
important so that large osmotic pressures are not created.
Significant osmotic pressure differences will cause fluid to
be gained or lost during microdialysis sampling (16). Fluid
loss is often undesirable for analytical as well as biological
reasons. From an analytical perspective, oftentimes the
analysis requires a set volume. For example, a liquid
chromatographic analysis may require 10 mL of sample
and a standard enzyme-linked immunosorbent assay
(ELISA) may require 100 mL of sample. From a biological
perspective, fluid loss can be undesirable in some tissues
that are particularly sensitive, such as the brain. Furthermore, brain tissue is also highly sensitive to ionic concentration alterations since such changes can alter
neurotransmitter release (17). By maintaining an osmotic
balance, the fundamental mass transport mechanism for
moving an analyte from the extracellular fluid space (ECF)
to the dialysate lumen is principally diffusion.
Probe Geometry
Microdialysis sampling is typically considered to be synonymous with the term intracranial dialysis sampling
because of its neuroscience origins. The first intracranial
dialysis device was a linear design that traversed longitudinally through different brain regions. A variety of
different probe designs have been described in many different review articles (1820). Microdialysis probe design
has evolved to allow use of the device for biomedical

applications beyond neuroscience. Linear geometry microdialysis sampling devices for neuroscience were not as
useful as the pushpull cannula that could be inserted into
known brain regions (e.g., striatum, hippocampus, substantia nigra) based on known stereotaxic coordinates that
in some cases are < 1 mm wide in rat brain. To overcome
this challenge for neurochemistry studies, more rigid cannula designs were created that can be inserted into specific
brain regions and are now commercially available from a
variety of sources (21).
As microdialysis sampling devices became a standard
tool used by neuroscientists, researchers in other fields
began to realize its great in vivo analysis potential. Principally, the use of the probes for collection of endogenous or
xenobiotic components in blood and peripheral tissues
became of interest (22). In these tissues, a rigid stainless
steel cannula causes tissue damage and may make awake
and freely moving experiments with animals quite difficult. Cannula designs using Teflon or fused silica are
commonly used for to make flexible probes for either sampling in soft peripheral tissues (e.g., skin or liver) or for
blood sampling. Linear probe designs have also been reintroduced after originally being applied to brain studies and
are now used for insertion into soft peripheral tissues. An
additional advantage of these flexible designs is they also
allow for studies in awake and freely moving animals in
peripheral tissues. Recent research interests in transgenic
mice have forced the creation of smaller microdialysis
sampling devices (23).
Probe Materials
Semipermeable hollow fiber membranes used for microdialysis sampling are the same as those used for kidney
dialysis. Different polymeric semipermeable membranes
have been used in microdialysis sampling probes and are
listed in Table 2. Typical materials include cellulose-based
membranes (cuprophan or cellulose acetate), polycarbonate/polyether blends, polyacrylonitrile, and polyethersulfone. These membranes span a wide range of molecular
weight cutoffs (MWCO) from 5000 to 100,000 Da. Choice of
the membrane to be used during microdialysis sampling
requires both analyte molecular weight information, as
well as where the probe will be implanted as some tissue
Table 2. Commercially Available Microdialysis Membrane

Dimensionsa
Outer radius, mm
Inner radius, mm
Wall thickness, mm
Molecular weight cutoff
Outer surface area, mm2
a
PC
PES
PAN
CUP
250
200
50
20,000
6.28
250
205
45
100,000
6.28
170
120
50
29,000
4.27
120
95
25
6,000
3.01
The data provided here is that given by the manufacturers of the microdialysis probes. It is not known if the radii are for dry or wet membranes.
The abbreviations are as follows: PC polyether/polycarbonate, PES
polyethersulfone, PAN polyacrylonitrile (or AN-69), CUP cuprophan.
CMA Microdialysis, Inc sells PC, PES, and CUP membranes. Bioanalytical
Systems, Inc sells PAN membrane probes.
403
regions (particularly in the brain) are too narrow for > 500
mm external diameter membranes.
Semipermeable hollow fiber dialysis membranes can be
obtained with a known molecular weight cut off (MWCO).
The MWCO can be experimentally determined for a hollow
fiber using several different experimental methods. The
primary method used to determine MWCO for hollow fiber
membranes is to continuously pass through the fiber
lumen over a long period of time (24 h or greater) a solution
containing known molecular weight markers. Known
solutes that are rejected by the membrane are then used
to calculate membrane MWCO. In practice, the MWCO is
really not an absolute number, but rather the median of a
range. This molecular weight rejection range is highly
dependent on the semipermeable membrane materials
pore distribution and can exhibit either a narrow or broad
MWCO range (24).
Originally, the purpose of microdialysis sampling was to
use the dialysis membrane as a means to provide a sample
for chemical analysis that did not require further sample
preparation steps such as protein removal. Intracranical
dialysis applications typically target hydrophilic analytes
with molecular weights < 500 Da. For these applications,
dialysis membranes with low MWCO of 50006000 Da
were commonly used to reject larger analytes and proteins
so to allow liquid chromatographic analysis without
further sample purification.
Recently, there has been a greater interest of applying
microdialysis sampling to collect peptides and proteins.
There have only been a few reports describing the use of
different types of dialysis membranes towards the collection of large molecules, such as peptides and proteins (25).
This is unfortunate as the types of commercially available
membranes that are capable of providing the performance
characteristics necessary for protein collection are relatively few. Kendrick extensively compared the recovery
performance of different amino acids and peptides among
different types of dialysis membranes (26). Torto et al.
compared the dialysis collection efficiency for a series of
saccharides [glucose (DP1), maltose (DP2), though maltoheptaose (D7)] among many different types of dialysis
membranes (polyamide, polyethersulfone, and polysulfone) as well as different MWCO between 6 and 100
kDa (27). In some cases, membranes with similar MWCO
and different chemistry exhibited similar recovery values.
Whereas, some of the polysulfone membranes with 100
kDa MWCO exhibited quite low recovery for these low
molecular weight analytes when compared to membranes
with similar chemistry, but lower MWCO.
A set of model proteins including insulin (5.7 kDa),
cytochrome c (12.4 kDa), ribonuclease A (13.7 kDa), lysozyme (14.4 kDa), and human serum albumin (67 kDa)
were tested with different polymeric membranes and
molecular weight cutoffs ranging between 20 and 150
kDa (28). All the membranes had similar external diameters (500 mm). Among the different membranes, only
the polyethersulfone (100 kDa MWCO) commercially
available from CMA Microdialysis, Inc and a Fresenius
polysulfone membrane (150 kDa MWCO) exhibited similar recovery characteristics for the above-mentioned set of
model proteins.
404
Membrane MWCO cannot be used as a means to specifically predict how well an analyte will be recovered
during a microdialysis sampling procedures. New microdialysis sampling practitioners sometimes mistakenly
believe that analytes near the membrane MWCO will be
recovered. Although a membrane with 100 kDa MWCO
allows some transport of molecules of this molecular
weight, the recovery of an analyte of this size will be
significantly < 1% (if at all) of the external sample concentration during microdialysis sampling. Dialysate analyte concentrations rarely reach equilibrium with the
external sample concentrations except under unique conditions (very low flow rates and long membranes). For
dialysate concentrations to reach those of the tissue medium surrounding the probe and thus approach equilibrium
with the surrounding tissue concentrations, low perfusion
fluid flow rates or long membranes are required in order to
achieve residence times sufficient to obtain equilibration.
During microdialysis sampling, the perfusion fluid only
passes once through the inner membrane lumen with
residence times on the order of seconds. Since this is in
contrast to the methods used to obtain membrane MWCO,
it is not surprising that sampled analyte molecular weight
range is reduced due to the perfusion fluid making only one
pass through the device. In general, the analyte molecular
weight that easily passes through the membrane with 10%
or greater recovery is roughly one tenth of the MWCO as
shown in Fig. 2 (29). However, this is not an absolute value
and different analytes have been reported to be difficult to
dialyze across particular membranes. With rare exception
(30), hydrophobic analytes typically are poorly dialyzed
during microdialysis sampling (31,32). Furthermore, AN69 membranes (polyacrylonitrile), which are sometimes
used for kidney dialysis and have been used for in-house
microdialysis probes, carry a negative charge that may
cause rejection of certain negatively charged analytes
(33,34).
Microdialysis sampling requires an inlet and outlet tube
to be attached to the membrane. The length and inner
diameter of the outlet tube attached to the membrane
Jv PD p Dp=l
Recovery, Delivery, and Localized Infusion

Sensor devices are highly specific analytical detectors and
can only be used to detect analytes that physically contact
100
Relative Recovery %
10
10
PC Membrane
0.1
200 400 600 800 1000 1200 1400
Molecular Weight
10
PAN Membrane
0.1
(1)
these hollow fiber semipermeable membranes is possible

and the extent of this ultrafiltration is related to the
volumetric flux (Jv) shown in Eq. 1, where P is the permeability coefficient for the membrane, l is the length across
the membrane (e.g., the membrane thickness), and Dp and
Dp the hydrostatic and osmotic pressure differences (35).
Different membranes have difference permeability coefficients. The physical manifestation of this fluid loss is that
the probe appears as if it is sweating during the dialysis
procedure and is often observed with larger MWCO membranes.
Peptides and proteins diffuse very slowly across the
small pores of membranes with MWCO between 5000
and 30,000 Da. To improve the relative recovery of these
analytes, larger 100 kDa dialysis membranes have become
commercially available. The disadvantage of these larger
MWCO membranes is that they often exhibit ultrafiltration due to their larger pore sizes. Ultrafiltration fluid
losses across 100 kDa or larger MWCO dialysis membranes
should be determined prior to in vivo experiments. In
particular, the ultrafiltration is exacerbated by the use
of long outlet tubing with narrow diameter (a common
need with awake and freely moving animal experiments).
Osmotic balancing agents, such as dextrans or albumin,
are commonly added to microdialysis perfusion fluids
passed through 100 kDa or larger MWCO membranes to
prevent excessive ultrafiltration as well as to prevent nonspecific adsorption on the device materials (36,37). While
these agents are passed through the membrane, their
potential loss to the surrounding tissue space has not been
reported.
100
Relative Recovery %
Relative Recovery %
100
affects membrane backpressure. For some hollow fiber

membranes, convective fluid loss (ultrafiltration) across
CUP Membrane
0.1
200 400 600 800 1000 1200 1400

Molecular Weight
200 400 600 800 1000 1200 1400

Molecular Weight
Figure 2. Semilog graph of relative recovery versus molecular weight for PC, PAN, and CUP
membranes using different perfusion fluid flow rates. Flow rates are 0.5 mLmin1 (&), 1.0 mLmin1
(), 2.0 mLmin1 (~) and 5.0 mLmin1 (~). Adapted from Ref. 29).
LUMEN
Q (L/min)
tative fraction of the analyte concentration in the surrounding ECF. Since most microdialysis conditions are
such that equilibrium between the dialysate and the sample is not obtained, a calibration has to be used to relate
dialysis concentrations to external sample concentrations.
Extraction efficiency (Ed) is used to relate the dialysis
concentration to the sample concentration. The steadystate Ed equation is
ECF
Dialysis
Membrane
Substrate + Reactant
Product
Substrate or
Inhibitor
Inhibitor
More Endogenous Analyte
Product
or
Endogenous
Analyte
405
Ed
Product
or
Endogenous
Analyte
Coutlet Cinlet
Ctissue, 1 Cinlet
(2)
1
1 exp
Qd Rd Rm Re Rt
Figure 3. Localized infusion. A substrate or drug is locally

infused through the microdialysis sampling probe. It diffuses
into the ECF and then either reacts to form a product or causes
a biochemical event to increase the concentrations of other
molecules.
the sensor. Microdialysis sampling provides extensive flexibility with respect to ways in which it can be applied.
Typical microdialysis sampling applications use what is
termed the recovery mode where the device is placed into a
sample matrix and analytes diffuse into the inner-fiber
lumen of the probe (Fig. 1). Alternatively, the device can be
used as a delivery device where a compound is infused
through the probe causing either some alteration in a
biochemical event (enzymatic reaction, enzymatic inhibition, or receptor binding) followed by collection of an endogenous analyte or enzymatic product. The probe can also be
used to pass a substrate for a chemical or biochemical
reaction and the products of that reaction can then be
locally sampled as shown in Fig. 3. Unlike a specific sensor,
microdialysis sampling devices allow for many physiological and biochemical processes to be studied within living
tissue. Different examples of this approach for a variety of
different tissues are shown in Table 3.
DEVICE CALIBRATION
Theoretical Foundations
Typical microdialysis sampling operating conditions (flow
rates between 0.5 and 2.0 mLmin1) will yield a represen-
shown below in Eq. 2, where Coutlet is the analyte concentration exiting the microdialysis probe, Cinlet is the analyte
concentration entering the microdialysis probe, Ctissue,1 is
the analyte tissue concentration far away from the probe,
Qd is the perfusion fluid flow rate and Rd, Rm, Re, and Rt are
a series of mass transport resistances for the dialysate,
membrane, external sample, and a trauma layer that
exists at the interface of the probe membrane and the
tissue as defined in Scheme 1 (48,49).
The resistance terms are additive and understanding
how these resistance terms affect Ed is vitally important
with respect to experimental design. Throughout the
microdialysis sampling process, collected analytes must
diffuse through at least three regions (tissue, membrane
and dialysate) in order to exit the microdialysis probe. Each
mass transport resistance term defined in Scheme 1 has a
diffusive component, which indicates that analytes with
smaller diffusion coefficients will exhibit much lower Ed.
The combined resistance contributions from Rd and Rm can
be experimentally determined in vitro by collecting dialysates at different flow rates. A plot of the natural log of
(1-Ed) versus 1/Qd should yield a straight line, which can be
regressed to determine the additive values for Rd and Rm.
In addition to this information, an in vitro Ed experiment
performed at 37 8C with stirring to cause the sample resistance, Re, to approach a zero value, will yield the highest
possible in vivo Ed.
The variables shown in Eq. 2 illustrate that a combination of perfusion fluid flow rate (Qd), as well as mass
transport resistances for the dialysate, membrane, and
tissue medium external to the microdialysis probe affect
Ed. Decreasing Qd allows for a greater fluid residence time
within the dialysis membrane thus allowing analyte
concentration to increase along the membrane axis.
Table 3. Some Examples of Localized Infusion Using Microdialysis Sampling

Type (substrate or
inhibitor)
Inhibitor
Substrate
Substrate
Substrate
Substrate and
inhibitor
Substrate
Infused Compound
Measured Analyte or Application
Tissue
Cocaine
Substance P and other neuropeptides
Salicylic acid or 4-hydroxybenzoic acid
Phenol or acetaminophen
Angiotensin, phosphoramidon, captopril
Dopamine
Proteolytic Products
2,3-DHBA, 2,5-DHBA, 3,4-DHBA
Metabolites
Metabolites and enzymatic inhibition
Brain
Brain
In Vitro, Brain
Liver
Renal Cortex
Suc-(Ala)3-pNA
Elastase (protease) activity
In Vitro
References
38
39,40
4143
44,45
46
47
406
Scheme 1. The multiple mass transport equations used to

describe microdialysis sampling. D is the diffusion coefficient
through the dialysate, Dd, membrane, Dm, and sample, Ds. The
parameter L is the membrane length; G (cm) is a composite
function; kep(r), km(r), and kc(r) are kinetic rate constants as a
function of radial position (r) from the microdialysis probe.
Additional term definitions can be found in Ref. 48.
Fig. 4 shows a typical Ed curve simulated using the above

equations (only Rd and Rm assuming a well-stirred system) for analytes with different aqueous diffusion coefficients. Fig. 4 clearly shows how microdialysis sampling
membranes even for a hypothetical case perform in a
manner that is consistent with diffusion being the major
contributor affecting recovery. This scenario for the diffusivity is especially true for protein collection as many
proteins of interest such as the cytokines have molecular
weight values that can begin to approach the molecular
weight cutoff limit for the dialysis membrane. In these
cases, the protein diameter can begin to approach the
values for the pore diameters resulting in restricted diffusion through the membrane, higher membrane mass
transport resistances and thus reduced analyte recovery.
The parameter Ed is highly dependent on several physiochemical parameters (analyte diffusion coefficient,
perfusion fluid flow rate, membrane pore size, and membrane surface area), kinetic uptake into cells (50) and the
microvasculature (51), as well as the overall ECF
volume fraction. The mass transport, resistances shown
in Scheme 1 include analyte diffusivity terms for all three
regions of mass transport, as well as kinetic terms for
1.0 E -5 cm /s (e.g., glucose)

2
1.0 E -6 cm /s (e.g., neuropeptide)
2
2.7 E -7 cm /s (e.g., TN F- )
100
Simulated Ed %
80
60
40
20
0
0
2
3
4
Perfusion Flow Rate (L/min)
Figure 4. Simulated Ed using Bungay et al. (48) model with the

following aqueous diffusion (Daq) coefficients and with a
membrane diffusion value of 0.2Daq with length of 10 mm and
Ri (200 mm) and Ro (250 mm).
the tissue space, as shown in the above equations. Tissue

diffusive and kinetic properties of the sample surrounding
an implanted microdialysis probe will dictate the how
reduced the in vivo Ed will be from the maximum possible
in vitro Ed value at any particular flow rate. For hydrophilic analytes, it is generally assumed they diffuse only in
the ECF that surrounds the tissue cellular components.
This ECF space comprises approximately 20% of the overall tissue volume (52). Typically hydrophilic analytes have
to diffuse around the cells en route to the microdialysis
probe, the overall effective diffusive path length is
increased due to the tortuous path traversed by the analyte. This tortuosity alters the tissue diffusion coefficient
which can be approximated using Decf Daq/l2, where l
has a value of 1.5. In addition to the alteration in
diffusive characteristics, tissues are vascularized and have
active cellular components. Depending on the analyte, the
active components will affect the overall microdialysis Ed.
In addition to these parameters influencing microdialysis Ed, analyte properties affect the shape and the time to
reach steady state for the concentration profile to the
microdialysis probe. Analytes that diffuse rapidly and have
a rapid supply to the tissue have narrow concentration
profiles to the dialysis probe. Conversely, analytes that
slowly diffuse and are not readily supplied to the tissue
space will have concentration profiles to the dialysis probe
that are not as steep. During microdialysis material is
removed from the sampling site and the extent to which
matter is removed is a function diffusive and kinetic parameters applied to that particular analyte, for example, how
rapidly it the analyte replenished to the ECF from either
capillaries (drugs) or cellular release processes. This has
been a concern by others particularly as it relates to the
understanding of dopamine transmission in the brain (53).
However, dopamine is a special case and has rapid release
and uptake kinetics in the ECF. For analytes that are
poorly transported across the capillary space in the brain
along with analytes that do not undergo significant uptake
(e.g., drugs), their relative recoveries are generally lower
than those with higher uptake/kinetic rates. It may appear
to be counterintuitive to think that analytes with very
rapid kinetic removal from the space surrounding the
microdialysis probe have increased relative recovery. However, the higher removal rates cause the concentration
profile to the dialysis probe to have a much greater gradient
to the device as compared to a poorly removed analyte
which would have a much shallower concentration profile
to the device. For analytes with similar ability to diffuse
through the membrane, that is, their membrane diffusion
coefficients are nearly equal, the flux should be greater for
the sharper concentration gradient thus causing greater
relative recovery.
The theoretical foundations for microdialysis sampling
during steady state operations derived by Bungay et al.
have been widely used to corroborate many different in vivo
experimental observations. In a series of papers focused on
neurotransmitters, Justices group has studied how uptake
inhibition decreases microdialysis Ed 38,54,55. Stenken
et al. 56 showed that since kinetic removal of targeted
analyte may in some cases be additive, the inhibition of a
particular cytochrome P450 isoform for phenacetin and
Calibration Methods
If the different variables shown in Eq. 2 and Scheme 1 are
known prior to experimentation, then it is possible to
predict the in vivo Ed. However, the difficulty with in vivo
experiments is that obtaining values for variables is highly
challenging. Furthermore, it is difficult to obtain an absolute value for Ctissue and an in vivo calibration requires that
Ctissue (Eq. 2) be known. Even today, many experiments
employing microdialysis sampling are semiquantitative
and dialysate concentrations obtained are approximations
of Ctissue at best. Since quite often ratios of dialysate
analyte concentrations are obtained before and after some
input (e.g., pharmacological or physical) with no attempt to
measure the Ctissue during the experiment. To date, the in
vivo calibration of implanted devices, including microdialysis probes, is an active area of research in the analytical
chemistry and bioengineering communities and many
authors have reviewed this subject (58,59). The principal
difficulty with respect to obtaining a reliable device calibration is the inability to fully reproduce in vitro all the
salient physiological features of tissue including permeation across capillaries and uptake processes (60,61).
Initial microdialysis sampling calibration focused on
using an in vitro Ed calculation to estimate tissue analyte
concentration. This approach gives a rough estimate of Ed
and may provide an incorrect calibration factor. Furthermore, in vitro methods used for Ed measurement are
affected by temperature (microdialysis sampling again is
inherently a diffusion separation method), as well as sample stirring. A well-stirred buffer medium provides a mass
transport external medium mass transport resistance (Re)
that approaches a value of zero. It is important to note that
a quiescent medium does provide diffusional mass transport resistance and thus relative recoveries performed in
vitro under stirred conditions will be different than those
performed using quiescent conditions (48). How close a
quiescently determined in vitro Ed is to the in vivo Ed
would be wholly dependent upon the tissue kinetic properties for the targeted analyte. In other words, an analyte,
such as dopamine, may exhibit higher in vivo Ed than in
vitro quiescent Ed due to its extensive uptake kinetics
causing a steeper concentration gradient to the dialysis
probe as compared to the in vitro quiescent Ed measurement. Differences in the ability of the analyte to diffuse
through the tissue space due to increased tortuosity and
decreased volume fraction led to empirical methods that
could be used to amend in vitro relative recovery calibration determinations (62). These methods focused on differences in tissue diffusion properties, but did not include the
role of kinetic affects on microdialysis Ed causing significant errors for estimating in vivo values for C tissue.
Jacobson et al. (63) were the first to try to create a more
analyte-specific calibration procedure for microdialysis
sampling. In their work, varying the perfusion fluid flow

rates through the dialysis probe derived an analyte-specific
membrane mass transport coefficient, K, shown below in
Eq. 3, where A is the membrane surface area and Qd is the
dialysate volumetric flow rate. Eq. 3
Coutlet
1 expKA=Qd
Ctissue
(3)
is similar to Eq. 2, showing how the model of Bungay et al.

incorporated previously known experimental results. In
this case, the product (-KA) is related to the sum of the
fraction of the mass transport resistance terms. Experimental results from this work immediately showed that
understanding the underlying in vivo mechanisms affecting microdialysis Ed was more complicated than initially
expected. These researchers found that different amino
acids exhibited different in vivo mass transport coefficients. This was unexpected since the amino acids would
be expected to have very similar diffusion coefficients due
to their similar molecular weight. This data began to lead
to the understanding that analyte properties (diffusion and
kinetics) in the tissue play a major role with respect to
microdialysis sampling calibration. An extension of calibration approach of Jacobson et al. is to pass the perfusion
fluid through the dialysis probe so slowly that zero flow is
approached and nearly 100% relative recovery as shown in
Fig. 5. In this case, the goal is to calculate Csample by
attempting to reach an equilibrium state across the microdialysis membrane (64).
The most widely used calibration method for microdialysis sampling is based on knowing that diffusive flux
should not occur across the dialysis membrane when the
analyte concentration inside the perfusion fluid matches
the concentration external to the microdialysis probe. This
method was originally demonstrated by Lo nnroth and has
been called by a variety of names including Lo nnroth plot,
100
80
Ed %
antipyrine metabolism did not significantly alter Ed. This

suggested that multiple kinetic components (capillary permeability plus metabolism) are important for analyte
removal from liver tissue. Elmquist has shown that transporter inhibition in brain causes alterations in Ed for
different drugs (57).
407
60
40
20
0
0
Flow Rate (L min1)

Figure 5. Mathematically modeled Ed for an approach to zero
flow.
408
Concentration (Cout - Cin )
20
10
0
Point of no-net-flux
10
20
0
50
100
150
Inlet Concentration (M)
200
Figure 6. Lonnroth plot using hypothetical data.
method of no net flux (NNF) and method of zero net flux

(ZNF). The ZNF method requires the tissue analyte concentration be at a steady state and has been used to
determine basal concentrations for many analytes. The
probe is perfused with different analyte concentrations
that are either above or below the targeted analyte concentration. With these different perfusion studies, the loss
or gain of analyte across the microdialysis probe can be
determined and then plotted versus the inlet concentration
as shown in Fig. 6. The analyte concentration is determined by the x-axis intercept and the relative recovery is
the absolute value of the calibration line slope. A major
drawback with this approach is the extensive amount of
research time that has to be invested during the different
perfusion studies. In particular, this is quite difficult to
achieve with exogenous analytes (drugs) that would
require a continuous infusion to achieve steady-state
concentrations.
Quite often what is desired from a microdialysis sampling procedure is the analyte concentration during some
sort of pharmacological challenge to an animal. This much
needed analyte temporal information is not possible to
collect with the requirement of steady state for the ZNF
method. To overcome this problem and to gain information
regarding the probe calibration from sample to sample,
internal standards have been used during microdialysis
sampling. Typically, internal standards have been chemicals with similar physicochemical properties as the targeted analyte. Some early work in internal standards
proposed using one standard, such as antipyrine or 3H
water to allow assessment of sample-to-sample differences
should they arise throughout the duration of the microdialysis sampling events (65). Antipyrine would be a suitable reference for probe-to-probe variability because of its
highly hydrophilic nature and ease of chemical detection.
Urea has also been used as a microdialysis calibration
reference for different metabolism studies (6668). It is
again important to note that the extraction efficiency of any
particular analyte is a combined function of the different
mass transport regions : dialysate, membrane, and most
importantly tissue. Most likely during an in vivo micro-
dialysis sampling experiment, the variability from sample

to sample will occur due to alterations in tissue physiology,
such as blood flow, metabolism, and uptake, which would
serve to alter the tissue resistance and thus Ed. For this
reason, it is generally preferred to have internal standards
with similar tissue diffusion and kinetic properties as the
analyte. Finding an appropriate internal standard is not a
trivial task since analytes with similar structure and diffusive properties may also compete for enzymatic sites and
may inhibit metabolic pathways that are important to
removal and thus Ed values. However, this possibility must
be considered in context of the tissue being sampled, as well
as other additive kinetic properties (e.g., uptake into cells
or capillaries) that may have a much greater impact on the
Ed. For example, in the brain, the kinetic process that has
been shown for several neurotransmitters to be most
weighted towards affecting Ed are the neuronal uptake
processes rather than metabolism processes. Additionally,
in the liver, it appears that capillary blood flow and permeability are the primary contributors toward the Ed value
obtained. Internal standards for peptides and proteins may
be much harder to devise as receptor binding or for the
cytokines, binding to the proteoglycan components of the
ECF space may affect Ed and thus using molecular weight
markers, such as inulin (69,70) or higher molecular weight
fluorescein-labeled dextrans (e.g., FITC-Dextran 3000,
FITC-Dextran 10,000) may only serve to report back diffusional mass transport differences during the duration of
microdialysis sampling.
Effect of Probe Insertion Trauma
Insertion of microdialysis probes causes tissue damage
(71,72). Although this has been known for quite some time,
it has generally been overlooked by many microdialysis
sampling users. The extent to which this insertion trauma
affects the integrity of the microdialysis sampling concentrations and its true overall importance has been debated
in the literature. The biomaterials literature is full of
descriptions of the cellular events that occur after a foreign
body implantation (73). It is known that edema occurs at
the site of probe implantation (74) along with the recruitment of polymorphonuclear leukocytes (75,76) and matrix
metalloproteinases (extracellular matrix remodeling
enzymes) (77). Moderately reduced analyte flux to microdialysis probe chronically implanted has been reported for
glucose (78).
The validity of the ZNF calibration methods for in vivo
calibration, as well as determination of Ctissue for some
analytes, has recently been a concern for neuroscientists
interested in dopamine. Many careful studies performed by
Michaels group illustrated that dopamine concentration
measurements obtained with microelectrodes and microdialysis sampling devices were quite different (79,80). In
particular, microdialysis sampling devices often exhibited
much lower basal concentrations of dopamine than microelectrodes. Additional concerns have been raised for drug
bloodbrain barrier studies (81,82). Between these two
examples, dopamine collection via microdialysis sampling
appears to be the most severely affected because of its
release and uptake sites being compromised due to the
insertion trauma (49,83,84). In essence, the creation of a

trauma layer creates four separate mass transport regions
during microdialysis sampling : the dialysate, membrane,
trauma layer, and normal tissue that need to be accounted
for during data interpretation.
ANALYSIS OF MICRODIALYSIS SAMPLES
Microdialysis sampling is essentially married to appropriate detection methods for the collected dialysates. In addition to providing a means to sample from an in vivo site,
microdialysis sampling also provides a relatively protein
free or clean sample for chemical analysis. The only selectivity imparted into a microdialysis membrane is its molecular weight cutoff. For this reason, as long as a targeted
analyte can diffuse through the membrane, the microdialysis sampling probe can be used as an in vivo chemical
collection device. Thus, assuming the targeted analyte can
pass through the dialysis membrane pores coupled with
the appropriate analytical methods, a microdialysis sampling device could be considered to be essentially an allpurpose in vivo sensor (85). There is an extensive literature
that has reviewed the associated analytical chemistry for
making measurements in microdialysis samples (86). Additional reviews include: Adell et al. (87), Chaurasia (88),
Church and Justice (89), Davies and Lunte (90), Horn (91),
Kennedy (92), Kennedy et al. (93). Lunte et al. (94), Lunte
and Lunte (95), Obrenovitch (96), Parkin et al. (97), and
Parrot et al. (98).
Sample Volume Limitations
The major bottleneck 25 years ago for microdialysis sampling gaining more wide-spread and universal acceptance
had to do with the analytical detection method sample
volume limitations. During the early stages of microdialysis sampling, the primary analytical detection methods
used for analyte quantification were liquid chromatography (LC) coupled with various types of detectors [ultravioletvisible (UVVis), fluorescence, and electrochemical]. In
addition to LC methods, radioimmunoassay (RIA) was
occasionally used for peptides and proteins. Twenty-five
years ago, it was not uncommon to require 2550 mL of
sample for LC analyses. Today, 50100 mL of sample is still
needed for standard immunoassays. The trade off that had
to occur became one of either obtaining higher concentration recovery across the membrane by using low perfusion
flow rates (1 mLmin1 or less) or gaining sufficient temporal resolution by going to faster flow rates to achieve
sufficient sample volumes for chemical analysis.
With the exception of glucose and lactate, many of the
endogenous as well as xenobiotic analytes sampled using
microdialysis had either micromolar (mM; 10-6 M) to nanomolar (nM 10-9 M) concentrations. These low concentrations often pushed the limitations of common analytical
equipment since for most analytes an approximate detection limit with most UV-Vis detectors is roughly in the low
mM range and for fluorescence and electrochemical detectors their detection limits are approximately in the nM
range. The need to be able perform analytical measurements from such low volume dialysates drove analytical
409
method development in multiple directions towards systems that could accommodate the low volumes without
sacrificing method sensitivity, as well as development of
high throughput methods that allowed for increased temporal resolution. Presently, there are many commercially
available technologies that allow for samples that are
<1 mL (e.g., capillary electrophoresis) or have duty cycles
that are <1 min.
Separations-Based Methods for Microdialysis Sample
Quantitation
Using separation methods, such as LC or capillary electrophoresis, for the quantitation of microdialysis samples is
highly advantageous since these methods can be quickly
adapted to many different analytes. Before the extensive
use of microdialysis sampling for studies of neurochemical
transmission, the use of in vivo voltammetry for analysis of
neurotransmitters was just beginning to be described as a
method for catecholamine (dopamine and norepinephrine)
(99,100). The difficulty with using these methods was that
electrode potentials needed to oxidize the catecholamines,
as well as their metabolites (3, 4-dihydroxyphenylacetic
acid, DOPAC) were similar. Furthermore, it was soon
discovered that during vesicular release of dopamine, very
high concentrations of ascorbic acid were released (101).
For this reason, in vivo voltammetry of these important
neurochemicals became more challenging since all of these
chemicals can be oxidized at or below the same potential.
The advantage of separations methods with appropriate
detectors is that components including targeted analytes,
as well as endogenous and exogenous interferences (see
Fig. 7) can be appropriately separated and quantified.
Thus, an additional advantage of using chromatographic
methods is that chromatographic methods provide intrinsic multiplexing capabilities for the chemical analysis of
microdialysis samples if different analytes are expected in
the same samples.
Liquid Chromatography. Liquid chromatographic methods have been used for analyzing a broad class of analytes
from microdialysis samples including catecholamines,
amino acids, pharmaceuticals, and their metabolites. Several articles are available that describe the necessary
requirements for microdialysis sample analysis using LC
(102,103). Liquid chromatographic separations methods
are well suited to microdialysis samples because of the
high salt content contained in the perfusion fluids. Salts
are generally not retained by the LC stationary phase and
are therefore eluted in the chromatographic void volume.
The resolving power of LC stationary phases allows for
multiple analytes to be quantified during a single chromatographic run. Different detectors have been applied to LC
separations for quantitation of dialysis samples.
Capillary Electrophoresis. Capillary electrophoresis
(CE) is a separation method that involves passing an
electric field across a micron-sized (25 to 75 mm internal
diameter) capillary so as to allow separation of analytes
based on their additive electrophoretic and electrosmotic
mobilities. Neutral components in capillary electrophoresis
410
H2
C
OH
C
H2
NH2
C
H H2
HO
NH2
HO
OH
OH
DOPAMINE
NOREPINEPHRINE
Figure 7. Example of different analytes
that can be oxidized at approximately the
same potential using a carbon electrode
under physiological conditions. The formal
potential for dopamine, norepinephrine, and
DOPAC are 0.7 V versus Ag/AgCl. For
acetominophen, the formal potential is
0.8 V versus Ag/AgCl. For ascorbic acid,
the oxidation formal potential is 0.15 V
versus Ag/AgCl.
H3C
NH
HOCH2
H
OH
O
H2
C COOH
O
HO
OH
ACETAMINOPHEN
(PARACETAMOL)
will elute based on their electroosmotic mobility. Compared

to LC analysis, CE provides greater separation efficiency
and the possibility of faster separations. An additional
advantage of CE is that an enormous research effort has
been placed into microfabrication of CE devices onto microchips. This suggests the possibility for point-of-care technologies to be integrated with microdialysis sampling.
Like LC, CE has been used for a wide range of microdialysis sampling analyses including catecholamine neurotransmitters, amino acid neurotransmitters, and
pharmaceutical compounds. Capillary electrophoresis
has one main disadvantage and that has to do with the
poor UV detection limits due to the path length being
significantly reduced. However, sensitive detection can
be achieved using electrochemical or laser-induced fluorescence detection approaches.
Although there are several disadvantages with CE
detection, that is, no standard equipment, requirement
of expertise, there is one advantage to this detection
method for microdialysis samples. By using pH-stacking
methods, significant on-column preconcentration (100 or
greater) can be achieved (104). The mismatch of pH
serves to greatly concentrate the sample zone on the
head of the capillary column. This allows for online
preconcentration to occur during the CE experiment.
For collection and detection of peptides or proteins
this preconcentration can serve to be highly useful.
An additional advantage of CE is that extremely fast
separations on the order of seconds can be achieved making this technology similar to that of a separations-based
biosensor (105).
Fast Separations. Microdialysis sampling can be a slow
temporal process due to the need to collect sufficient sample with sufficient relative recovery. To make microdialysis
sampling more sensor like in terms of its response time, a
number of groups have worked on achieving high speed
separations, as well as direct coupling dialysate outflow the
detection method. This is particularly important for studies of neurotransmitter dynamics. While electrochemical
approaches for studies of catecholamine neurotransmis-
HO
OH
ASCORBIC ACID
OH
3,4-DIHYDROXYPHENYLACETIC ACID
(DOPAC)
sion provide millisecond time resolution, microdialysis

sampling is hindered by the turn-around time for the
analytical method. High speed detection capability of cycle
times of <1 min have been reported by many different
groups using both liquid chromatographic, as well as
capillary electrophoresis separations including 1 s time
resolution with neurotransmitters (106) and <1 min resolution with pharmacokinetic analyses (107).
Examples of Different Types of Detection
Electrochemical Detection. Liquid chromatography
coupled with electrochemical detection (LCEC) is both a
highly sensitive and selective method for analysis of compounds that can undergo an electrochemical reaction (oxidation or reduction). In this sense, LCEC is highly suited
to the chemical analysis of important biogenic amines
(dopamine, norepinephrine, etc) obtained from microdialysis probes implanted into the brain (108). For these
measurements, the electrochemical detector has a glassy
carbon electrode that allows for oxidation of the amines at
potentials of roughly 700 mV versus a Ag/AgCl reference
electrode. The LCEC excels in this analysis task because
these analytes can be oxidized and furthermore their basal
concentrations are in the low to mid-nanomolar range. The
advantage of the separation is evident when considering
that catecholamine metabolites (DOPAC, HVA) have basal
concentrations that are 1001000 times greater than the
clinically relevant catecholamines (dopamine, norepinephrine, serotonin).
In addition to catecholamine detection of microdialysis
samples, LCEC analysis has been applied to low concentration analytes obtained under varying conditions of oxidative stress. In these cases, typically either salicylic acid
or 4-hydroxybenzoic acid is directly infused through the
microdialysis probe to locally deliver a trapping agent as
shown in Fig. 3 (109). These benzoic acids react with
hydroxyl radical to form catechols which can then be
separated and detected by LCEC (110). The LCEC has
also been used to quantify the DNA oxidative damage
biomarker, 8-dOHdGuanosine (111,112).
Electrochemical detection can be made to be more selective by altering the electrode surface. Gold electrodes
coated with Hg to create an amalgam are highly selective
toward thiols such as cysteine and glutathione with low
potentials needed for oxidation 150 mV versus Ag/AgCl
(113). Lunte and OShea used this approach for glutathione
detection using CE (114).
In addition to electrode modification, packed enzyme
beds containing specific oxidase can be used prior to electrochemical detection. This is commonly applied to detection of the neurotransmitters choline and acetylcholine.
Acetylcholine and choline can be separated chromatographically and then an enzymatic bed containing acetylcholine oxidase and choline oxidase is placed at the end of the
column. These specific enzymatic reactions produce hydrogen peroxide which is then detected downstream at a
platinum electrode (115). Note that more recent developments have attempted to immobilize the enzymes specifically to the electrode (116).
Examples of Fluorescence for Dialysates. Fluorescence
detection is often employed when a known derivatization
method can be applied to dialysate samples to improve
method detection limits or to create a molecule that has a
better handle for detection. For microdialysis samples,
fluorescence derivatization is commonly applied to important amino acid neurotransmitters such as glutamate and
GABA (117119) and occasionally to biogenic amines (e.g.,
dopamine and norepinephrine) (120).
Examples of Mass Spectrometry for Dialysates. Mass
spectrometry (MS) is a unique LC detector in that as long
as the analyte has the ability to form an ion, MS can be used
for analysis. However, mass spectrometric detection can be
difficult with microdialysis samples because of the high salt
content. This method has been particularly useful with
neuropeptides because of their low concentrations. A problem with mass spectrometric detection is that salts from
the dialysis perfusion fluid can cause analyte ionization
suppression that leads to dramatically decreased detection
capability for the method (121). Salts from dialysates are
often removed via a column-switching technique that preconcentrates the analyte onto a C18 phase followed by
desorption and detection (122). However, the use of nanoelectrospray devices can also reduce some of the problems
associated with salt adducts (123125). A particularly
powerful method of LC-MS has been the ability to perform
ionization in stages, which allows for structural elucidation
of unknowns. The Kennedy group has been particularly
successful with this approach for sequencing neuropeptides obtained from microdialysis samples (39,126).
Sensor Attachment to Microdialysis Probes
The microdialysis sampling process results in a relatively
analytically clean (little to no protein) sample. Separations
methods provide extensive analysis flexibility because they
can be quickly optimized to the targeted analytes. However, there are in vivo monitoring situations where only
one or a few analytes are targeted and highly specific
analysis methods are available. A particular case in point
411
is the continuous detection of glucose or lactate from

diabetic humans (127). The primary advantage of coupling
a sensing device to the end of the microdialysis sampling
device is the sample matrix is simply saline passing across
the analytical sensor. This prevents many of the difficulties
associated with biofouling of implanted sensors (128).
However, a critical problem for glucose sensing using this
approach is that it can only provide information regarding
the glucose concentration fluctuations throughout the day,
but cannot really serve as an alarm system because of the
lag times that are 2030 min as compared to normal
glucose sensors of a few minutes (129). Despite this concern, there is great value in using specialized sensors to a
microdialysis device because of the clean sample delivered
to the sensing device.
The use of biosensors attached to the end of microdialysis probes has become highly useful for clinical neuroscience where measuring glucose and lactate and in
some cases other neurotransmitters are needed to understand homeostatic mechanisms (130,131). Most biosensors
attached to dialysis probes have been for glucose, lactate, or
glutamate detection (132134). Cook has published an
interesting approach combining immunoassay with electrochemical detection for specific measurements of coritsol
(135).
Immunoassay for Peptide and Protein Detection
Peptide and protein analysis of microdialysis samples is
challenging since the concentrations of these targeted
analytes are often in the ngmL1 to pgmL1 levels. This
requires either highly sensitive fluorescence derivatization techniques for use with capillary electrophoresis (136)
or sensitive immunoassays. Conventional immunoassays
require 50100 mL of sample. To obtain these volumes
requires the use of high flow rates (2 mLmin1 or greater)
or very long collection times. In most cases, because
highly sensitive radioimmunoassays (RIA) are used,
higher flow rates are used to achieve moderate temporal
resolution.
It is becoming increasingly evident that cellular communication in biological systems is highly complex and
networked. Despite the tremendous growth in microdialysis sampling to monitor cellular biochemistry and an
increased interest in peptide and protein detection in
dialysates, there has been relatively little research towards
new analytical methods that can detect peptides and proteins in low volume dialysate samples. A few approaches
have been published that require 80100 mL of sample for
detection of several different proteins (137,128). Multiplexed assays (up to 25 analytes or more) that can be
performed on a single sample have been recently created
for immunology studies of the important inflammatory
mediator class of cytokine proteins.
Highly sensitive multiplexed immunoassay platforms
based on particle-based flow cytometry has become commercially available that allows cytokine measurements in
50-mL sample volumes (139,140). The limit of detection for
these assays fall into the low pg/mL range comparable and
have been compared and validated against standard
ELISA methods (141). The use of these particle-based
412
immunoassays is highly advantageous to the samplelimited microdialysis process and the sample volume
needed has been decreased to < 25 mL by our group for
cytokine detection. The advantage of these bead-based
immunoassays for microdialysis samples is that several
analytes can be analyzed in a single low volume sample. To
illustrate the significant advantage that the bead-based
immunoassay provides, if six separate cytokines were to be
quantified in microdialysis samples using standard ELISA
techniques more than 600 mL of sample would be needed.
Using a flow rate of 1 mLmin1, this would require 10 h of
microdialysis sampling.
Mass versus Concentration Recovery
Microdialysis sampling Ed is a concentration recovery term
and Ed increases as fluid flows decrease through the dialysis fiber creating longer residence times. Conversely,
overall mass recovery typically increases as the flow rate
increases as shown in Fig. 8. For some analytical applications, this increase in mass recovery may prove to be highly
beneficial since it opens up possibilities for analytical preconcentration methods for the increased dialysate
volumes.
MICRODIALYSIS SAMPLING APPLICATIONS
Microdialysis sampling applications have now been widely
used in many different mammalian species including
humans. The applications in humans have included studies in cancer, dermatology, immunology, pharmacokinetics and neuroscience. Many of these applications
have been extensively reviewed by others and therefore
will not be extensively discussed here. It is again important
to note that microdialysis sampling has to date been principally applied to applications in neuroscience for the past
1 105 cm2s1 e.g., glucose

1 106 cm2s1 . e.g., neuropeptide
2.7 106 cm2s1 e.g., TNF-alpha
Simulated Mass Recovery Normalized
1.0
0.8
0.12
0.11
0.10
0.6
0.09
0.08
0.5
0.07
0.06
0.4
0.05
0.04
0.3
0.03
0
0.2
0.1
0.0
0
Neuroscience Applications
Microdialysis sampling has been in the neuroscientist toolbox for > 25 years. This device has been principally applied
to neurotransmitter collection. Early on, the primary focus
was in rat and more recently with probe redesigns and the
biomedical value of knockouts, additional studies have
been performed in mice (23,142). Current research interests focus on bridging the gap between animal models and
human studies.
Reviewing all the microdialysis literature for neuroscience applications is a daunting task since the microdialysis sampling technique is now in wide use. Some of the
applications have already been mentioned in the Analysis
section of this article. However, several reviews and a book
(see Bibliography section) are available as background.
These reviews have covered general aspects of neurochemical collection with microdialysis sampling (143146),
microchemical analysis (147,148), and controversial
aspects of neurotransmitter collection (84,149151)
Neuropeptides. With successful sampling of hydrophilic neurotransmitters with microdialysis sampling, the
next logical analyte class to target was neuropeptides. Like
neurotransmitters, the quantitation of neuropeptides is
challenging with microdialysis sampling due to their low
concentrations. Furthermore, their lower diffusion coefficients cause their Ed values to be low. Temporal resolution
can also be an issue since quite often immunoassays that
require 50100 mL are often used for detection. Despite
these limitations, many neuropeptides have been sampled
using microdialysis sampling and have been reviewed > 15
years ago (152,153). With the increased use of mass spectrometry for neuropeptide detection of dialysates (154,155)
coupled with additional bead-based immunoassays, the
application space for microdialysis sampling of neuropeptides should increase tremendously.
Pharmacokinetics
0.9
0.7
three decades. However, over the past decade, more microdialysis sampling applications in other areas are now being
described.
Volumetric Flow Rate (L min1)

Figure 8. Modeled microdialysis sampling mass recovery for
different analytes with different diffusion coefficients.
Microdialysis sampling has been applied for pharmacokinetic studies in animals and humans. The great advantage
here is that microdialysis sampling tremendously
decreases the overall number of animals used for a pharmacokinetics study. Typical pharmacokinetic studies in
rodents require the animal to be sacrificed at each time
point used for the analysis. Microdialysis sampling allows
for collection throughout the time course of the experiment
because it can be easily inserted into the jugular vein.
Again, because of the highly flexible nature of liquid chromatographic analysis for drug studies, the microdialysis
sampling technique can be rapidly applied to new drugs
and their metabolites.
Microdialysis sampling applications in pharmacokinetics have been extensively reviewed. In addition to general reviews of the subject (156158), there have been
reviews focused on data analysis (159,160) and calibration
(161,162). One of the more important points to consider
when working with pharmacokinetic data obtained using

microdialysis sampling is that a microdialysis sample
represents a concentration average over the collection time
period. This is in contrast to blood sampling that represents
the analyte concentration directly at the sample time.
Clinical Applications
Clinical applications of microdialysis sampling have grown
tremendously over the past decade and will continue to
grow (163). At the present time, microdialysis sampling
methods in human subjects have focused on studies in
peripheral tissues for glucose collection (164167) or drug
distribution (168170). Additional applications have been
to determine gut barrier dysfunction (171). Microdialysis
sampling applications in tumors has spanned both pharmacokinetic investigations (172174), as well as collection
of growth factors and cytokines (175). Microdialysis sampling has also been applied to human brain studies that
have focused on understanding the underlying altered
biochemistry that occurs when head trauma to drug distribution (176180). Microdialysis sampling has been used
as a means to monitor pharmacokinetics in the human
dermis and has been compared to blister suction techniques (181). While both sampling methods produced similar
data, it was found that microdialysis sampling was much
easier to handle for both the patient and clinician.
Monitoring different growth factors and cytokines is
becoming more important in clinical medicine since these
proteins are known to affect cell-to-cell signaling and communication and are therefore becoming important biomarkers to measure. In particular, a group of proteins that are
of great in vivo interest are the cytokines. Cytokines are
potent, transient, and highly localized soluble messenger
proteins (680 kDa) produced by T-cells and macrophages that control nearly every aspect of the immune
system (182). Cytokines exhibit complex interactions and
therefore it is often more important to determine the
concentration and cytokine profile after an immune challenge rather than the concentration of one single cytokine.
Microdialysis sampling is an ideal technique to achieve
real time in situ monitoring of these important protein
mediators and also has been recently described for proteomics applications (183). The application of microdialysis to
this area is now emerging as potential approach for clinical
413
in vivo studies in both healthy and diseased subjects to

recover targeted cytokine molecules from the exact action
sites and has recently been reviewed by Clough (25).
Commercially available microdialysis probes with a 100
kDa MWCO membrane have been used for in vivo microdialysis of some cytokines (184,185). It is important to note
that microdialysis sampling provides localized sampling
and thus insight into localized concentrations of cytokines
that cannot be achieved via sampling from blood plasma.
This has been recently demonstrated with the cytokine IL6 where its interstitial fluid concentration was 100-fold
higher than that found in the plasma (186).
Cytokines have low Ed through 100 kDa membranes. To
improve cytokine Ed larger MWCO membranes (3000 kDa)
typically used for plasmaphoresis have been used (187).
Others are beginning to use the 3000 kDa MWCO membrane for collection of IL-6 and TGF- b1 (186,188190). The
range of in vitro recoveries for different cytokine proteins is
shown in Table 4.
EVALUATION AND FUTURE USE
Microdialysis sampling has become a mature technology
for neurotransmitter collection and pharmacokinetic
determinations in animals. Clinical microdialysis sampling applications provide the greatest opportunity for
growth. Despite the extensive biomedical use of conventional microdialysis sampling, there are still aspects of the
device that could be tremendously improved.
As currently practiced, microdialysis sampling in animals can be cumbersome due to the tubing lines required.
Work in Luntes group has focused on making micropumps
using osmotic pumps as means to create line-free dialysis
device (191,192). Reducing the microdialysis size by creating it on a microchip also has some advantages given that a
decreased volume flow chamber may allow rapid equilibration across the device allowing Ed to approach nearly 100%
(193195).
A common problem with microdialysis sampling is
the difficulty incurred when sampling hydrophobic
analytes. This is an area with great promise with respect
to either new device development or improvements to
existing microdialysis sampling technology. Albumin
is commonly included in the perfusion fluid to block
Table 4. Cytokine In Vitro Relative Recovery and Relevant Physicochemical Propertiesa

Cytokine
Ed% 0.5 mLmin1
Ed% 1.0 mLmin1
IL-2
IL-4
IL-5
IL-6
IL-10
IL-12p70
IFN-g
MCP-1
TNF-a
4.5
2.9 (12)
12.0
6.5 (12)
1.3
1.0 (12)
4.8
1.4 (6)
Not performed
N.D.
2.0
1.4 (18)
24.5
4.8 (6)
8.0
2.9 (15)
2.9
1.1 (15)
7.5
2.6 (15)
1.0
0.5 (15)
2.7
0.8 (6)
1.1
0.3 (3)
N.D.
1.5
0.8 (21)
13.1
3.9 (6)
4.3
1.1 (18)
MW, kDa
Conformation
Active Protein, kDa
17.2
13.6
13.1
21.7
18.8
35 & 40
15.9
13.1
17.3
Monomer
Monomer
Homodimer
Monomer
Homodimer
Heterodimer
Homodimer
Homodimer
Homotrimer
17.2
13.6
26.2
21.7
37.6
75
31.8
26.2
51.9
Cytokine standards were either 1250 or 2500 pgmL1. All solutions were quiescent at room temperature. A CMA/20 10-mm 100-kDa PES membrane was used
for these studies performed in our laboratory. The numbers in parentheses after the RR values are the number of trials (n). All data are reported as mean
SD.
N.D. is not detected.
414
Membrane
Q,
(Lmin1)
Sample
SA
S+A
dC L
dx
Analyte (S)
2pDmem (Co C L ) p Ri kC L
Q ln( Ro / Ri )
Q
At x = 0, C = C o
At x = L, C = C L
Figure 9. Schematic of microdialysis enhanced transport. The

analyte (S) or substrate for an affinity agent (A) binds with the
affinity agent in the perfusion fluid. Affinity agents include
antibodies, cyclodextrins or other supramolecular agents. The
symbols in the equation are as follows: Co Concentration at
x 0, CL Concentration at x L (membrane length), Dmem
membrane diffusion coefficient, k first-order rate constant for
the reaction between the analyte, S, and the affinity agent, A, Q
perfusion fluid flow rate and Ri, Ro inner and outer radii.
nonspecific adsorption sites within the microdialysis polymeric materials (196,197), and is still widely practiced
(198). Unfortunately, this adds protein back into the dialysis perfusion that for small analytes causes difficulties
with CE and LC analyses. A second approach involves
using lipids, such as Intralipid, a material used to suspend
hydrophobic drugs, to coat the nonspecific adsorption sites
on the dialysis membrane or tubing (196,199). It is not
known how compatible this approach is for CE or LC
analysis.
Cyclodextrins have been used as perfusion-fluid additives for improving the microdialysis recovery for different
analytes (200,201). A general scheme for the enhancement
process is shown in Fig. 9. Cyclodextrins are well-known
cyclic oligosacharides that have the capability to form
inclusion complexes with various organic molecules by
the capture of the guest molecule into a hydrophobic
central cavity (202). Dialysate samples that contain cyclodextrin can be injected into an LC column without alteration to the chromatographic separation parameters, for
example, plate number and peak width (203). Cyclodextrins are not selective, which for this enhancement
approach is beneficial given that they can be applied to a
wide variety of low molecular weight hydrophobic organic
molecules (204,205). However, a difficulty with this
approach is that cyclodextrins can diffuse out of the dialysis membrane that may complicate some applications. An
additional difficulty is that cyclodextrins can interfere with
affinity-based detection assays (immunoassays) due to
competition between the analyte and cyclodextrin versus
analyte and an antibody.
To overcome the free diffusion of cyclodextrin out of the
dialysis tubing, different type of solid supports have been
used for enhancement. Markides group has described the
use of a solid-vehicle support for improving neuropeptide
microdialysis relative recovery coupled with LCMS detection (206,207). More specific enhancements can be
achieved using antibody-immobilized beads for flow cytometry (208). With the antibody-enhancement approach,
increases in microdialysis sampling relative recovery of
412-fold were achieved for different cytokines.
Microdialysis sampling is already beginning to be used
in metabolomic studies using LCMS detection methods
(209). With the creation of microcoil nuclear magnetic
resorance (NMR) that can measure nanoliter samples
(210,211) the detection possibilities for dialysate samples
are greatly increased. This approach has been recently
applied to metabolomic studies with microdialysis sampling in brain (212).
The combination of these new discoveries for microdialysis sampling shows the enormous potential for solving
many clinical biomedical problems with this device. The
future for microdialysis sampling and device spin-offs is
quite bright. New developments in instrumentation and
collection procedures will provide much clinical benefit.
Microdialysis sampling has moved from a sampling
device exclusively used to collect low molecular weight
hydrophilic neurotransmitters to applications requiring
the collection of larger proteins. Many cellular signaling
processes occur via small molecule (nitric oxide, norepinephrine, acetylcholine, eicosanoids, etc.), peptide (angiotensin, etc.) as well as proteins (cytokines). Cytokine
profiling for different disease states has become quite
important. For the applications of both neuropeptide and
larger protein collection such as the cytokines, the principal limitation to collecting these molecules is the diffusion
properties of these analytes through the dialysis membrane. Proteins are difficult to collect through dialysis
membranes since they can exhibit both nonspecific adsorption to the polymeric materials as well as diffusion restrictions through the membrane pores. Despite the variety of
applications of microdialysis sampling to medical and
scientific research, there have been few new developments
with respect to improving the overall sampling efficiency
for difficult to dialyze samples. As clinical proteomics and
biomarker collection becomes better understood for making clinical predictions, it will be necessary to have sampling methods that can meet these clinical needs.
ACKNOWLEDGMENTS
Microdialysis sampling research in the Stenken laboratory
is currently funded by NIH EB001441. The support from
the John Wiley & Sons, Inc. editorial staff is greatly
appreciated.
BIBLIOGRAPHY
1. Frangioni JV. In vivo near-infrared fluorescence imaging.
Curr Opin Chem Biol 2003;7:626634.
2. Ntziachristos V, Tung CH, Bremer C, Weissleder R. Fluorescence molecular tomography resolves protease activity in vivo.
Nature Med 2002;8:757761.
3. Sevick-Muraca EM, et al. Near-infrared imaging with fluorescent contrast agents. Handbook Biomed Fluoresc 2003; 445
527.
4. Myers RD, Adell A, Lankford MF. Simultaneous comparison
of cerebral dialysis and pushpull perfusion in the brain of
5.
6.
7.
8.
9.
10.
11.
12.
13.
rats: a critical review. Neurosci Biobehav Rev 1998;22:371

387.
Ellmerer M, et al. Measurement of interstitial albumin in
human skeletal muscle and adipose tissue by open-flow microperfusion. Am J Physiol 2000;278:E352E356.
Haaverstad R, Romslo I, Larsen S, Myhre HO. Protein concentration of subcutaneous interstitial fluid in the human leg.
A comparison between the wick technique and the blister
suction technique. Inter J Microcirculation, Clin Exper
1996;16:111117.
Wiig H, Sibley L, DeCarlo M, Renkin EM. Sampling interstitial fluid from rat skeletal muscles by intermuscular wicks.
Am J Physiol 1991;261:H155H15.
Nedrebo T, et al. (2004). Differential cytokine response in
interstitial fluid in skin and serum during experimental
inflammation in rats. J Physiol 2004;556:193202.
Bito LZ, et al. Concentration of free amino acids and other
electrolytes in cerebrospinal fluid, in vivo dialyzate of brain, and
blood plasma of the dog. J Neurochem 1966;13:10571067.
Delgado JM, et al. Dialytrode for long term intracerebral
perfusion in awake monkeys. Arch Inter Pharmacodyn Ther
1972;198:921.
Ungerstedt U, Pycock C. Functional correlates of dopamine
neurotransmission. Bull Schweizerischen Akad Medizinischen
Wisse 1974;30:4455.
Plock N, Kloft C. Microdialysis - theoretical background and
recent implementation in applied life-sciences. Eur J Pharmaceut Sci 2005;25:124.
Davies MI, et al. Analytical considerations for microdialysis
sampling. Adv Drug Deliv Rev 2000;45:169188.
14. Benveniste H, Huttemeier C. Microdialysis-theory and application. Prog Neurobiol 1990;35:195215.

15. de Lange ECM, de Boer AG, Breimer DD. Methodological
issues in microdialysis sampling for pharmacokinetic studies.
Ad Drug Del Rev 2000;45:125148.
16. Borg N, Stahle L. Recovery as a function of the osmolality of the
perfusion medium in microdialysis experiments. Anal Chim
Acta 1998;375:135141.
17. Moghaddum B, Bunney BS. Ionic compositions of microdialysis
perfusing solution alters the pharmacological responsiveness
and basal outlflow of striatal dopamine. J Neurochem
1989;53:652654.
18. Davies MI. A review of microdialysis sampling for pharmacokinetic applicaitons. Analyt Chim Acta 1999;379:227249.
19. De Lange CMA, De Boer AG, Breimer DD. Intracerebral
microdialysis. Intro BloodBrain Barrier 1998; 94112.
20. Bourne JA (2003). Intracerebral microdialysis: 30 years as a
tool for the neuroscientist. Clin Exp Pharmacol Physiol
2003;30:1624.
21. North American and International Commercial Sources of
Microdialysis Probes. Available at CMA Microdialysis Inc.
www.microdialysis.se; Bioanalytical Systems Inc., www. bioanalytical.com; SciPro www.scipro.com .
22. De la Pena A, Liu P, Derendorf H. Microdialysis in peripheral
tissues. Adv Drug Deli Rev 2000;45:189216.
23. Boschi G, Scherrmann JM. Microdialysis in mice for
drug delivery research. Adv Drug Deliv Rev 2000;45:271281.
24. Mulder M. Basic Principles of Membrane Technology. Dordrecht, The Netherlands: Kluwer Academic Publishers 1991.
25. Clough G. Microdialysis sampling of large molecules. The
AAPS Journal E-Pub May 16, 2005.
26. Kendrick KM. Use of microdialysis in neuroendocrinology.
Methods Enzymol 1989;168:182205.
415
27. Torto N, et al. Optimal membrane choice for microdialysis

sampling of oligosaccharides. J Chromatog A 1998;806:265
278.
28. Kjellstrom S, et al. Microdialysisa membrane based sampling technique for quantitative determination of proteins.
Chromatogr 1999;50:539546.
29. Snyder KL, et al. Diffusion and calibration properties of
microdialysis sampling membranes in biological media. Analyst 2001;126:12611268.
30. Schuck VJA, Rinas I, Derendorf H. In vitro microdialysis
sampling of docetaxel. J Pharmaceutic Biomed Anal 2004;36:
807813.
31. Groth L, Jrgensen A. In vitro microdialysis of hydrophilic
and lipophilic compounds. Analyt Chim Acta 1997;355:7583.
32. Mary S, et al. Assessment of the recovery of three lipophilic
psoralens by microdialysis: An in vitro study. Inter J Pharmaceut 1998;161:713.
33. Patterson SL, Sluka KA, Arnold MA. A novel transverse
pushpull microprobe: in vitro characterization and in
vivo demonstration of the enzymatic production of adenosine in the spinal cord dorsal horn. J Neurochem
2001;76:234246.
34. Patterson SL, Sluka KA, Arnold MA. A novel transverse
pushpull microprobe: In vitro characterization and in vivo
demonstration of the enzymatic production of adenosine in
the spinal cord dorsal horn. [Erratum to document cited in A].
J Neurochem 2001;76:1955.
35. Ho WSW, Sikar KK, editors. Membrane Handbook New
York; Van Nostrand Reinhold: 1992.
36. Rosdahl H, Hamrin K, Ungerstedt U, Henriksson J, A microdialysis method for the in situ investigation of the action of
large peptide molecules in human skeletal muscle: Detection
of local metabolic effects of insulin. Intern J Biol Macromole
2000;28:6973.
37. Hamrin K, Rosdahl H, Ungerstedt U, Henriksson J. Microdialysis in human skeletal muscle: Effects of adding a colloid
to the perfusate. J Appl Physiol 2002;92:385393.
38. Smith AD, Justice JB. The effect of inhibition of synthesis,
release, metabolism and uptake on the microdialysis extraction fraction of dopamine. J Neurosci Methods 1994;54:75
82.
39. Haskins WE, et al. Capillary LC-MS2 at the attomole level for
monitoring and discovering endogenous peptides in microdialysis samples collected in vivo. Analy Chem 2001;73:5005
5014.
40. Freed AL, Cooper JD, Davies MI, Lunte SM. Investigation of
the metabolism of substance P in rat striatum by microdialysis sampling and capillary electrophoresis with laserinduced fluorescence detection. J Neurosc Methods
2001;109:2329.
41. Ste-Mari L, Boismenu D, Vachon L, Montgomery J. Evaluation of sodium 4-hydroxybenzoate as an hydroxyl radical trap
using gas chromatography-mass spectrometry and high-performance liquid chromatography with electrochemical detection. Ana Bioch 1996;241:6774.
42. Chen R, Stenken JA. An in vitro hydroxyl radical generation
assay for microdialysis sampling calibration. Anal Biochem
2002;306:4049.
43. Marklund N, Clausen F, Lewander T, Hillered L. Monitoring of reactive oxygen species production after traumatic
brain injury in rats with microdialysis and the 4-hydroxybenzoic acid trapping method. J Neurotrauma 2001;18:
12171227.
44. Scott DO, Lunte CE In vivo microdialysis sampling in the
bile, blood, and liver of rats to study the disposition of phenol.
Pharmaceut Res 1993;10:335342.
416
45. Stenken JA, Stathle L, Lunte CE, Southard MZ. Monitoring

in situ liver metabolism in rats using microdialysis. Comparison of a microdialysis mass-transport model predictions to
experimental metabolite generation data. J Pharmaceut Sci
1998;87:311320.
46. Kajiro T, Nakajima Y, Fukushima T, Imai K. A method to
evaluate the renin-angiotensin system in rat renal cortex
using a microdialysis technique combined with HPLC-fluorescence detection. Anal Chem 2002;74:45194525.
47. Steuerwald AJ, Villeneuve JD, Sun L, Stenken JA. In vitro
characterization of an in situ, microdialysis sampling assay
for elastase activity detection. J Pharmaceut Biome Analy. In
press.
48. Bungay PM, Morrison PF, Dedrick RL. Steady-state theory
for quantitative microdialysis of solutes and water in vivo and
in vitro. Life Sci 1990;46:105119.
49. Bungay PM, et al. Microdialysis of dopamine interpreted
with quantitative model incorporating probe implantation
trauma. J Neurochem 2003;86:932946.
50. Justice JB. Quantitative microdialysis of neurotransmitters.
J Neurosc Methods 1993;48:263276.
51. Clough GF, et al. Effects of blood flow on the in vivo recovery
of a small diffusible molecule by microdialysis in human skin.
J Pharmacol Exp Therapeut 2002;302:681686.
52. Nicholson C. Diffusion and related transport mechanisms in
brain tissue. Rep Prog Phys 2001;64:815884.
53. Newton AP, Justice JB. Temporal response of microdialysis
probes to local perfusion of dopamine and cocaine followed
with one-minute sampling. Analy Chem 1994;66:1468
1472.
54. Cosford RJO, Vinson AP, Kukoyi S, Justice JBJ. Quantitative
microdialysis of serotonin and norepinephrine: Pharmacological influences on in vivo extraction fraction. J Neurosci
Methods 1996;68:3947.
55. Vinson PN, Justice JB. Effect of neostigmine on concentration and extraction fraction of acetylcholine using quantitative microdialysis. J Neurosci Methods 1997;73:6167.
56. Stenken JA, Lunte CE, Southard MZ, Stahle L. Factors that
influence microdialysis recovery. Comparison of experimental and theoretical microdialysis recoveries in rat liver. J
Pharmaceut Sci 1997;86:958966.
57. Dai H, Elmquist WF. Drug transport studies using quantitative microdialysis. Methods Mol Med 2003;89:249264.
58. Stenken JA. Methods and issues in microdialysis calibration.
Anal Chim Acta 1999;379:337357.
59. Chen KC, et al. Theory relating in vitro and in vivo microdialysis with one or two probes. J Neurochem 2002;81:108
121.
60. Rice ME, Nicholson C. Diffusion and ion shifts in the brain
extracellular microenvironment and their relevance for voltammetric measurements. The brain is not a beaker: In vivo
vs. In vitro voltammetry. In: Boulton A, Baker G, Adams RN,
editors. Neuromethods, Vol. 27, Voltammetric Methods in
Brain Systems, ed. New York: Humana Press Inc.; 1995. pp
2779.
61. Reach G, Wilson GS. Can continuous glucose monitoring be
used for the treatment of diabetes? Analy Chem 1992;64:
381A386A.
62. Benveniste H, Huttemeir C. Microdialysis-theory and application. Progr Neurobiol 1990;35:195215.
63. Jacobson I, Sandberg M, Hamberger A. Mass transfer in
brain dialysis devicesa new method for the estimation of
extracellular amino acids concentration. J Neurosci Methods
1985;15: 263268.
64. Menacherry S, Hubert W, Justice JB. In vivo calibration of
microdialysis probes for exogenous compounds. Analyt Chem
1992;64:577583.
65. Yokel RA, Allen DD, Burgio DE. McNamara JP. Antipyrine
as a dialyzable reference to correct differences in efficiency
among and within sampling devices during in vivo
microdialysis. J Pharmacol Toxicol Methods 1992;27:135
142.
66. Strindberg L, Lonnroth P. Validation of an endogenous
reference technique for the calibration of microdialysis catheters. Scand J Clin Lab Investigation 2000;60:205
212.
67. Ronne-Engstrom E, et al. Intracerebral microdialysis in neurointensive care: The use of urea as an endogenous reference
compound. J Neurosur 2001;94:397402.
68. Ettinger SN, et al. Urea as a recovery marker for quantitative
assessment of tumor interstitial solutes with microdialysis.
Cancer Res 2001;61:79647970.
69. Leypoldt JK, Burkart JM. Small-solute and middle-molecule
clearances during continuous flow peritoneal dialysis. Adv
Peritoneal Dialysis 2002;18:2631.
70. Krejcie TC, et al. Modifications of blood volume alter the
disposition of markers of blood volume, extracellular fluid,
and total body water. J Pharmacol Exp Therap 1999;291:
13081316.
71. Benveniste H, Diemer NH. Cellular reactions to implantation
of a microdialysis tube in the rat hippocampus. Acta Neuropathol 1987;74:234238.
72. Grabb MC, et al. Neurochemical and morphological responses to acutely and chronically implanted brain microdialysis probes. J Neurosci Methods 1998;82:2534.
73. Anderson JM. Biological responses to materials. Ann Rev
Mater Res 2001;31:81110.
74. Dykstra KH, et al. Quantitative examination of tissue concentration profiles associated with microdialysis. J Neurochem 1992;58:931940.
75. Davies MI, Lunte CE. Microdialysis sampling for hepatic
metabolism studies: impact of microdialysis probe design
and implantation technique on liver tissue. Drug Metabolism
Disposition 1995;23:10721079.
76. Clapp-Lilly KL, et al. An ultrastructural analysis of tissue
surrounding a microdialysis probe. J Neurosci Methods,
1999;90:129142.
77. Planas AM, et al. Certain forms of matrix metalloproteinase9 accumulate in the extracellular space after microdialysis
probe implantation and middle cerebral artery occlusion/
reperfusion. J Cerebral Blood Flow Metabolism 2002;22:
918925.
78. Wisniewski N, et al. Analyte flux through chronically
implanted subcutaneous polyamide membranes differs in
humans and rats. Am J Physiol, Endocrinol Metabolism
2002;282:E1316E1323.
79. Lu Y, Peters JL, Michael AC. Direct comparison of the
response of voltammetry and microdialysis to electrically
evoked release of striatal dopamine. J Neurochem
1998;70:584593.
80. Borland LM, Shi G, Yang H, Michael AC. Voltammetric study
of extracellular dopamine near microdialysis probes acutely
implanted in the striatum of the anesthetized rat. J Neurosci
Methods 2005;146:149158.
81. Morgan ME, Singhal D, Anderson BD. Quantitative assessment of blood-brain barrier damage during microdialysis. J
Pharmacol Exp Therap 1996;277:11671176.
82. Groothuis DR, et al. Changes in blood-brain barrier permeability associated with insertion of brain cannulas and microdialysis probes. Brain Res 1998;803:218230.
83. Chen KC. Preferentially impaired neurotransmitter release
sites not their discreteness compromise the validity of
microdialysis zero-net-flux method. J Neurochem 2005;92:
2945.
84. Khan SA, Michael AC. Invasive consequences of using microelectrodes and microdialysis probes in the brain. TrAC,
Trends Anal Chem 2003;22:503508.
85. Ballerstadt R, Schultz JS. Sensor methods for use with
microdialysis and ultrafiltration. Adv Drug Del Rev 1996;
21:225238.
86. Davies MI, et al. Analytical considerations for microdialysis
sampling. Adv Drug Del Rev 2000;45:169188.
87. Adell A, Artigas F. In vivo brain microdialysis: Principles and
applications. Neuromethods 1998;32:133.
88. Chaurasia CS. In vivo microdialysis sampling: Theory and
applications. Biomed Chromatog 1999;13:317332.
89. Church WH, Justice Jr JB. On-line small-bore chromatography for neurochemical analysis in the brain. Adv Chromatog
1989;28:165194.
90. Davies IM, Lunte CE. Microdialysis sampling coupled on-line
to microseparation techniques. Chem Soc Rev 1997;26:215
222.
91. Horn TFW, Engelmann M. In vivo microdialysis for nonapeptides in rat braina practical guide. Methods 2001;23:41
53.
92. Kennedy RT. Bioanalytical applications of fast capillary electrophoresis. Anal Chim Acta 1999;400:163180.
93. Kennedy RT, et al. In vivo neurochemical monitoring by
microdialysis and capillary separations. Curr Opin Chem
Biol 2002;6:659665.
94. Lunte CE, Scott DO, Kissinger PT. Sampling living systems
using microdialysis probes. Anal Chem, 1991;63:773A780A.
95. Lunte SM, Lunte CE. Microdialysis sampling for pharmacological studies: HPLC and CE analysis. In: Brown PR,
Grushka E, editors. Advances in Chromatography. New
York: Marcel Dekker; 1996. pp 383432.
96. Obrenovitch TP, Zilkha E. Microdialysis coupled to online
enzymatic assays. Methods 2001;23:6371.
97. Parkin MC, Hopwood SE, Boutelle MG, Strong AJ. Resolving
dynamic changes in brain metabolism using biosensors
and on-line microdialysis. TrAC, Trends Analy Chem
2003;22:487497.
98. Parrot S, et al. Microdialysis monitoring of catecholamines
and excitatory amino acids in the rat and mouse brain:
Recent developments based on capillary electrophoresis with
laser-induced fluorescence detection-A mini-review. Cellular
Mol Neurobiol 2003;23:793804.
99. Wightman RM, Strope E, Plotsky PM, Adams RN. Monitoring of transmitter metabolites by voltammetry in cerebrospinal fluid following neural path stimulation. Nature London
1976;262:145146.
100. Adams RN. Probing brain chemistry with electroanalytical
techniques. Anal Chem 1976;48:1126A1138A.
101. Nagy G, Rice ME, Adams RN. A new type of enzyme electrode:
The ascorbic acid eliminator electrode. Life Sci 1982; 31:2611
2616.
102. Wages SA, Church WH, Justice Jr JB. Sampling considerations for on-line microbore liquid chromatography of brain
dialysate. Analy Chem 1986;58:16491656.
103. Kennedy RT, German I, Thompson JE, Witowski SR. Fast
analytical-scale separations by capillary electrophoresis and
liquid chromatography. Chem Rev 1999;99:30813131.
104. Arnett SD, Lunte CE. Investigation of the mechanism of pHmediated stacking of anions for the analysis of physiological
samples by capillary electrophoresis. Electrophoresis
2003;24:17451752.
105. Davies M, et al. Studies on animal to instrument hyphenation: Development of separation-based sensors for near realtime monitoring of drugs and neurotransmitters. Sample
Preparation for Hyphenated Analyt Tech 2004; 191220.
417
106. Rossell S, Gonzalez LE, Hernandez L. One-second time

resolution brain microdialysis in fully awake rats. Protocol
for the collection, separation and sorting of nanoliter dialyzate volumes. J Chromatogr B: Analy Technol Biomed Life
Sci 2003;784:385393.
107. Chen A, Lunte CE. Microdialysis sampling coupled on-line
to fast microbore chromatography. J Chromatog Analy Appl
1995;691:2935.
108. Kissinger PT, Refshauge C, Dreiling R, Adams RN. Electrochemical detector for liquid chromatography with picogram
sensitivity. Analyt Lett 1973;6:465477.
109. Acworth IN, Bogdanov MB, McCabe DR, Beal MF. Estimation of hydroxyl free radical levels in vivo based on liquid
chromatography with electrochemical detection. Methods
Enzymol 1999;300:297313.
110. Acworth IN, Bailey BA, Maher TJ. The use of HPLC with
electrochemical detection to monitor reactive oxygen and
nitrogen species, markers of oxidative damage and antioxidants: application to the neurosciences. Prog HPLC-HPCE
1998;7:356.
111. Bogdanov MB, et al. A carbon column-based liquid chromatography electrochemical approach to routine 8-hydroxy-2deoxyguanosine measurements in urine and other biologic
matrices: A one-year evaluation of methods. Free Rad Biol
Med 1999;27:647666.
112. Bogdanov MB, et al. Increased oxidative damage to DNA in
a transgenic mouse model of Huntingtons disease. J Neurochem 2001;79:12461249.
113. Stenken JA, Puckett DL, Lunte SM, Lunte CE. Detection of
N-acetylcysteine, cysteine and their disulfides in urine by
liquid chromatography with a dual-electrode amperometric
detector. J Pharmaceut Biomed Analysis 1990;8:8589.
114. Lunte SM, OShea TJ. Pharmaceutical and biomedical
applications of capillary electrophoresis/electrochemistry.
Electrophoresis 1994;15:7986.
115. Zackheim JA, Abercrombie ED. HPLC/EC detection and
quantification of acetylcholine in dialysates. Methods Mol
Med 2003;79:433441.
116. Kehr J, Dechent P, Kato T, Ogren SO. Simultaneous determination of acetylcholine, choline and physostigmine in
microdialysis samples from rat hippocampus by microbore
liquid chromatography/electrochemistry on peroxidase
redox polymer coated electrodes. J Neurosci Methods
1998;83:143150.
117. Kehr J. Determination of glutamate and aspartate in microdialysis samples by reversed-phase column liquid chromatography with fluorescence and electrochemical detection. J
Chromatog B: Biomed Sci App 1998;708:2738.
118. Kehr J. Determination of g-aminobutyric acid in microdialysis samples by microbore column liquid chromatography
and fluorescence detection. J Chromatog B: Biomed Sci App
1998;708:4954.
119. Tao L, Thompson JE, Kennedy RT. Optically gated capillary
electrophoresis of o-phthalaldehyde/b-mercaptoethanol
derivatives of amino acids for chemical monitoring. Analy
Chem 1998;70:40154022.
120. Yamaguchi M, et al. Determination of norepinephrine in
microdialysis samples by microbore column liquid chromatography with fluorescence detection following derivatization with benzylamine. Analy Biochem 1999;270: 296
302.
121. Mann M, Fenn JB. Electrospray mass spectrometry: principles and methods. Mass Spectrom 1992;1:135.
122. Emmett MR, Andren PE, Caprioli RM. Specific molecular
mass detection of endogenously released neuropeptides
using in vivo microdialysis/mass spectrometry. J Neurosci
Methods 1995;62:141147.
418
123. Wilm M, Mann M. Analytical properties of the nanoelectrospray ion source. Analy Chem 1996;68:18.
124. Juraschek R, Dulcks T, Karas M. Nanoelectrospray-more
than just a minimized-flow electrospray ionization source. J
Am Soc Mass Spectrom 1999;10:300308.
125. Gangl ET, Annan M, Spooner N, Vouros P. Reduction of
signal suppression effects in ESI-MS using a nanosplitting
device. Analy Chem 2001;73:56355644.
126. Haskins WE, et al. Discovery and neurochemical screening of
peptides in brain extracellular fluid by chemical analysis of in
vivo microdialysis samples. Analy Chem 2004;76:55235533.
127. Heinemann L. Continuous glucose monitoring by means
of the microdialysis technique: Underlying fundamental
aspects. Diabetes Technol Therap 2003;5:545561.
128. Wisniewski N, Moussy F, Reichert WM. Characterization of
implantable biosensor membrane fouling. Fresenius J
Analy Chem 2000;366:611621.
129. Ward WK. Subcutaneous glucose monitoring: Microdialysis
vs. intracorporeal. Diabetes Care 2002;25:410411.
130. ONeill RD, Lowry JP, Mas M. Monitoring brain chemistry
in vivo: Voltammetric techniques, sensors, and behavioral
applications. Crit Rev Neurobiol 1998;12:69127.
131. Jones DA, et al. On-line monitoring in neurointensive care
Enzyme-based electrochemical assay for simultaneous, continuous monitoring of glucose and lactate from critical care
patients. J Electroanalyt Chem, 2002; 538539, 243252.
132. Boutelle MG, Fellows LK, Cook C. Enzyme packed bed
system for the on-line measurement of glucose, glutamate,
and lactate in brain microdialysis. Analy Chem 1992;64:
17901794.
133. Volpe G, Moscone D, Compagnone D, Palleschi G. In vivo
continuous monitoring of ***L-lactate coupling subcutaneous microdialysis and an electrochemical biocell. Sensors
Actuators B 1995; 2425 138141.
134. Ryan MR, Lowry JP, ONeill RD. Biosensor for neurotransmitter L-glutamic acid designed for efficient use of L-glutamate oxidase and effective rejection of interference.
Analyst 1997;122:14191424.
135. Cook CJ. Real-time measurement of corticosteroids in conscious animals using an antibody-based electrode. Nature
Biotechnol 1997;15:467471.
136. Sandberg M, Weber SG. Techniques for neuropeptide determination. TrAC, Trends Analy Chem 2003;22:522527.
137. OConnor KA, et al. A method for measuring multiple
cytokines from small samples. Brain, Behavior, Immunity
2004; 18:274280.
138. Li Y, Schutte RJ, Abu-Shakra A, Reichert WM. Protein
array method for assessing in vitro biomaterial-induced
cytokine expression. Biomaterials 2005;26:10811085.
139. Vignali DAA. Multiplexed particle-based flow cytometric
assays. J Immunol Methods 2000;243:243255.
140. Kellar KL, et al. Multiplexed fluorescent bead-based immunoassays for quantitation of human cytokines in serum and
culture supernatants. Cytometry 2001;45:2736.
141. Kellar KL, Iannone MA. Multiplexed microsphere-based
flow cytometric assays. Exp Hemato 2002;30:12271237.
142. Xie R, Hammarlund-Udenaes M, De Boer AG, De Lange
ECM. The role of P-glycoprotein in blood-brain barrier
transport of morphine: Transcortical microdialysis studies
in mdr1a (/) and mdr1a (/) mice. Br J Pharmacol 1999;
128:563568.
143. Ungerstedt U. Introduction to intracerebral microdialysis.
Tech Behav Neural Sci 1991;7:322.
144. Sharp T, Zetterstrom T. In vivo measurement of monoamine
neurotransmitter release using brain microdialysis. Monit.
Neuronal Act 1992; 147179.
145. Westerink BHC, Timmerman W. Do neurotransmitters

sampled by brain microdialysis reflect functional release?
Anal Chim Acta 1999;379:263274.
146. Salamone JD. The behavioral neurochem of motivation:
methodological and conceptual issues in studies of the
dynamic activity of nucleus accumbens dopamine. J Neurosc Methods 1996;64:137149.
147. Justice Jr JB. Microchemical analysis in the brain. Microchem J 1986;34:1114.
148. Kennedy RT, et al. In vivo neurochemical monitoring by
microdialysis and capillary separations. Curr Opin Chem
Biol 2002;6:659665.
149. Fuxe K, Ferre S, Zoli M, Agnati LF. Integrated events in
central dopamine transmission as analyzed at multiple
levels. Evidence for intramembrane adenosine A2A/dopamine D2 and adenosine A1/dopamine D1 receptor interactions in the basal ganglia. Brain Res Rev 1998;26:258
273.
150. Del Arco A, Segovia G, Fuxe K, Mora F. Changes in dialysate
concentrations of glutamate and GABA in the brain: An
index of volume transmission mediated actions? J Neurochem 2003;85:2333.
151. Di Chiara G, Tanda G, Carboni E. Estimation of in-vivo
neurotransmitter release by brain microdialysis: The issue
of validity. Behav Pharmacol 1996;7:640657.
152. Kendrick KM. Use of microdialysis in neuroendocrinology.
153. Kendrick KM. Microdialysis measurement of in vivo neuropeptide release. J Neurosc Methods 1990;34:3546.
154. Andren PE, Lin-S. N, Caprioli RM. Microdialysis/mass
spectrometry. Mass Spectrom 1994;2:237254.
155. Andren PE, Farmer TB, Klintenberg R. Endogenous release
and metabolism of neuropeptides utilizing in vivo microdialysis microelectrospray mass spectrometry. Mass Spectrom Hyphenated Tech Neuropeptide Res 2002; 193213.
156. Stahle L, Microdialysis in pharmacokinetics. Eur J Drug
Metabol Pharmacokine 1993;18:8996.
157. de Lange ECM, Danhof M, de Boer AG. Breimer DD.
Methodological considerations of intracerebral microdialysis in pharmacokinetic studies on drug transport across the
blood-brain barrier. Brain Res Rev 1997;25:2749.
158. Hansen DK, et al. Pharmacokinetic and metabolism studies
using microdialysis sampling. J Pharmaceut Sci 1999;88:
1427.
159. Stahle L. Pharmacokinetic estimations from microdialysis
data. Eur J Clin Pharmacol 1992;43:289294.
160. Stahle L. Zero and first moment area estimation from
microdialysis data. Eur J Clin Pharmacol 1993;45:477481.
161. Hammarlund-Udenaes M, Paalzow LK. de Lange ECM.
Drug equilibration across the blood-brain barrierpharmacokinetic considerations based on the microdialysis method.
Pharmaceut Res 1997;14:128134.
162. Bungay PM, Dedrick RL, Fox E, Balis FM. Probe calibration
in transient microdialysis in vivo. Pharmaceut Res 2001;18:
361366.
163. Muller M. Science, medicine, and the future: Microdialysis.
Br Med J 2000;324:588591.
164. De Boer J, Korf J, Plijter-Groendijk H. In vivo monitoring of
lactate and glucose with microdialysis and enzyme reactors
in intensive care medicine. Inter J Artif Organs
1994;17:163170.
165. Weintjes KJ, et al. Microdialysis of glucose in subcutaneous
adipose tissue up to 3 weeks in healthy volunteers. Diabetes
Care 1998;21:14811488.
166. Gudbjoernsdottir S, et al. Direct measurements of the permeability surface area for insulin and glucose in human
167.
168.
169.
170.
171.
172.
173.
174.
175.
176.
177.
178.
179.
180.
181.
182.
183.
184.
185.
186.
187.
skeletal muscle. J Clin Endocrino Metab 2003;88:4559

4564.
Lonnroth P. Microdialysis in adipose tissue and skeletal
muscle. Hormone Metabol Res 1997;29:344346.
Blochl-Daum B, et al. Measurement of extracellular fluid
carboplatin kinetics in melanoma metastases with microdialysis. Br J Cancer 1996;73:920924.
Muller M, et al. In vivo drug-response measurements in
target tissues by microdialysis. Clin Pharmacol Therapeut
1997;62:165170.
Lindberger M, Tomson T. Stahle L. Validation of microdialysis sampling for subcutaneous extracellular valproic acid in
humans. Therapeut Drug Monitoring 1998;20:358362.
Solligard E, et al. Gut barrier dysfunction as detected by
intestinal luminal microdialysis. Int Care Med 2004;30:
11881194.
Chu J, Gallo JM. Application of microdialysis to characterize drug disposition in tumors. Adv Drug Del Rev
2000;15:243253.
Mader RM, et al. Penetration of capecitabine and its metabolites into malignant and healthy tissues of patients with
advanced breast cancer. Br J Cancer 2003;88:782787.
Johansen MJ, Thapar N, Newman RA, Madden T. Use of
microdialysis to study platinum anticancer agent pharmacokinetics in preclinical models. J Expe Therap Oncol 2002;
2:163173.
Dabrosin C. Microdialysis an in vivo technique for studies
of growth factors in breast cancer. Frontiers Biosci 2005;10:
13291335.
Hamani C, Luer MS, Dujovny M. Microdialysis in the
human brain: review of its applications. Neurolog Res
1997;19:281288.
Vespa P, et al. Increase in extracellular glutamate caused by
reduced cerebral perfusion pressure and seizures after
human traumatic brain injury: A microdialysis study. J
Neurosur 1998;89:971982.
Sherwin AL. Neuroactive amino acids in focally epileptic
human brain: A review. Neurochem Res 1999;24:1385
1395.
Bradberry CW. Applications of microdialysis methodology
in nonhuman primates: Practice and rationale. Crit Rev
Neurobiol 2000;14:143163.
Laruelle M. Imaging synaptic neurotransmission with in
vivo binding competition techniques: A critical review. J
Cerebral Blood Flow Metab 2000;20:423451.
Benfeldt E, Serup J, Menne T. Microdialysis vs. suction
blister technique for in vivo sampling of pharmacokinetics
in the human dermis. Acta Dermato-Venereolog 1999;79:
338342.
Lefkowitz DL, Lefkowitz SS. (2001) Macrophage-neutrophil
interaction: A paradigm for chronic inflammation revisited.
Immunol Cell Biol 2001;79:502506.
Maurer MH, et al. The proteome of human brain microdialysate. Proteome Sci 2003;1:715.
Sjogren F, Svensson C, Anderson C. Technical prerequisites
for in vivo microdialysis determination of interleukin-6 in
human dermis. Br J Dermatol 2002;146:375382.
Ao X, Rotundo RF, Loegering DJ, Stenken JA. In vivo
microdialysis sampling of cytokines produced in mice given
bacterial lipopolysaccharide. J Microbiol Methods
2005;62:327336.
Sopasakis VR, et al. High local concentrations and effects on
differentiation implicate interleukin-6 as a paracrine regulator. Obesity Res 2004;12:454460.
Winter CD, et al. A microdialysis method for the recovery of
IL-1beta, IL-6 and nerve growth factor from human brain in
vivo. J Neurosc Methods 2002;119:4550.
419
188. Rosendal L, et al. Increase in interstitial interleukin-6 of

human skeletal muscle with repetitive low-force exercise. J
Appl Physiol 2005;98:477481.
189. Heinemeier K, Langberg H, Olesen JL, Kjaer M. Role of
TGF-beta1 in relation to exercise-induced type I collagen
synthesis in human tendinous tissue. J Appl Physiol
2003;95:23902397.
190. Langberg H, Olesen JL, Gemmer C, Kjaer M. Substantial
elevation of interleukin-6 concentration in peritendinous
tissue, in contrast to muscle, following prolonged exercise
in humans. J Physiol 2002;542:985990.
191. Lin Y-C, Hesketh PJ, Lunte SM, Wilson GS. A micromachined diaphragm micropump. ProcElectrochem Soci
1995; 9527. 6772.
192. Hesketh PJ, et al. Biosensors and microfluidic systems.
Tribology Issues and Opportunities in MEMS, Proceedings
of the NSF/AFOSR/ASME Workshop on Tribiology Issues
and Opportunities in MEMS, Columbus, Ohio, Nov. 911,
1997, 1998; pp 8594.
193. Zahn JD, Trebotich D, Liepmann D. Microfabricated microdialysis microneedles for continuous medical monitoring.
Proceedings of the 1st Annual International IEEE/EMBS
Special Topics Conference on Microtechnologies in Medicine
& Biology. October 1214, 2000, Lyon, France; 2000 pp 375
380.
194. Talbot D, Liepmann D, Pisano AP. Microfabricated polysilicon microneedles for minimally invasive biomedical
devices. Biomed Microdevices 2000;2:295303.
195. Bergveld P, et al. Microdialysis based lab-on-a chip, applying a generic MEMS technology. Comprehen Analy Chem
2003;39:625663.
196. Carneheim C, Stahle L. Microdialysis of lipophilic compounds: a methodological study. Pharmacol Toxicol 1991;
69:378380.
197. Mueller M, et al. In vivo characterization of transdermal
drug transport by microdialysis. J Controlled Release
1995;37: 4957.
198. Trickler WJ, Miller DW. Use of osmotic agents in microdialysis studies to improve the recovery of macromolecules.
J Pharmaceut Sci 2003;92:14191427.
199. Kurosaki Y, Nakamura S, Shiojiri Y, Kawasaki H. Lipomicrodialysis: a new microdialysis method for studying the
pharmacokinetics of lipophilic substances. Biolog Pharmaceut Bull. 1998;21:194196.
200. Khramov AN, Stenken JA. Enhanced microdialysis extraction efficiency of ibuprofen in vitro by facilitated transport
with b-cyclodextrin. Analy Chem 1999;71:12571264.
201. Kjellstrom S, et al. Online coupling of microdialysis sampling with liquid chromatography for the determination of
peptide and non-peptide leukotrienes. J Chromatog A
1998;823:489496.
202. Rekharsky MV, Inoue Y. Complexation thermodynamics of
cyclodextrins. Chem Rev 1998;98:18751917.
203. Stenken JA, Chen R, Yuan X. Influence of geometry and
equilibrium chemistry on relative recovery during enhanced
microdialysis. Anal Chim Acta 2001;436:2129.
204. Khramov AN, Stenken JA. Enhanced microdialysis recovery of some tricyclic antidepressants and structurally
related drugs by cyclodextrin-mediated transport. Analyst
1999; 124:10271033.
205. Ward KW, et al. Enhancement of in vitro and in vivo
microdialysis recovery of SB-265123 using intralipid and
encapsin as perfusates. Biopharmaceut Drug Disposition
2003;24:1725.
206. Pettersson A, Markides K, Bergquist J. Enhanced microdialysis of neuropeptides. Acta Biochim Polon 2001;48:
11171120.
420
MICROFLUIDICS
207. Pettersson A, et al. A feasibility study of solid supported

enhanced microdialysis. Analy Chem 2004;76:16781682.
208. Ao X, Sellati TJ, Stenken JA. Enhanced microdialysis relative recovery of inflammatory cytokines using antibodycoated microspheres analyzed by flow cytometry. Anal
Chem 2004;76:37773784.
209. Kissinger CB, Kissinger PT. Can preclinical ADMET-PK
now be done more efficiently and effectively? Preclinica
2004;2: 319323.
210. Olson DL, Lacey ME, Sweedler JV. High-resolution microcoil NMR for analysis of mass-limited, nanoliter samples.
Anal Chem 1998;70:645650.
211. Wolters AM, Jayawickrama DA, Sweedler JV. Microscale
NMR. Curr Opin Chem Biol 2002;6:711716.
212. Khandelwal P, et al. Studying rat brain Neurochem using
nanoprobe NMR spectroscopy: A metabonomics approach.
Analy Chem 2004;76:41234127.
Further Reading
The most comprehensive sources for microdialysis sampling are
the book and the two separate journal issues shown below.
Robinson T, Justice JB, editors. Microdialysis in the Neurosciences. Amsterdam (The Nethernand): Elsevier; 1991.
Lunte CE Anal Chim Acta 1999;379:227369.
Elmquist WF, Sawchuk RJ. Microdialysis sampling in drug delivery. Adv Drug Del Res 2000;45:123307.
See also ELECTROPHORESIS;
GLUCOSE SENSORS; HYDROCEPHALUS, TOOLS
FOR DIAGNOSIS AND TREATMENT OF; PHARMACOKINETICS AND PHARMACODYNAMICS.
MICROFLUIDICS
GLENN M. WALKER
North Carolina State University
Raleigh, North Carolina
INTRODUCTION
Microfluidics is the study and application of fluids at the
microscale. The most common definition of the microscale
is that one or more device dimension be in the range of
11000 mm. For reference, the diameter of an average
human head hair is 150 mm, the average thickness of a
human fingernail is 360 mm, and the diameter of a human
red blood cell is 7 mm. Miniaturization technology, originally developed by the microelectronics industry, has
been used to create microscale fluid components and complete microfluidic systems with pumps, valves, and filters,
incorporated onto single microchips have been demonstrated.
By applying the analogy of the microelectronics industry (i.e., continuously incorporating more features into
smaller areas) a logical application of microfluidics is to
create lab-on-a-chip (LOC) systems. Lab-on-a-chip systems, also known as micro-total-analysis systems (mTAS),
incorporate the functionality of biology or chemistry
laboratories onto a single microfabricated chip. Ideally, a
LOC system would be able to execute all of the tasks
routinely performed in a biology or chemistry laboratory,
such as sample preconditioning, mixing, reaction, separa-
tion, and analysis. Labor- and time-intensive procedures

would be reduced to instant results derived from a series of
automated steps performed on a LOC.
Microscale fluid handling confers many advantages over
traditional lab operations (1). First, fluid quantities ranging from picoliters to microliters are used, thus reducing
the amount of sample required for tests. Second, the
amount of time required to perform some analyses (e.g.,
capillary electrophoresis) is reduced to seconds, which
means analyses can be conducted many times faster than
with traditional methods. Third, devices can be manufactured using microfabrication technology, which translates
into reduced cost per device; disposable LOC systems can
easily be envisioned.
In general, microfluidic devices are in early stages of
development and are most often found in academic
research laboratories. However, the benefits of these systems have been exploited to develop new medical devices
for clinical diagnostics and point-of-care testing. Commercial examples of devices that make use of LOC concepts are
discussed at the end of this article.
THEORY
Fluid Mechanics
The term microfluidics encompasses both liquid and gas
behavior at the microscale, even though in most applications the working fluid is a liquid. All of the concepts
discussed here are directed toward liquids. Other works
are available which provide information on gas behavior at
the microscale (2).
Fluid behavior at the microscale is different from that
commonly observed in everyday experiences at the macroscale, owing primarily to the very low Reynolds (Re) numbers of the flow regime plus the large surface area/volume
(SAV) ratios of the flow domain. As a consequence, viscous
forces and surface tension effects become dominant over
fluid inertia, and transport phenomena are purely diffusive.
Fluid flow at the microscale is typically laminar. Fluid
flows are classified based on their flow regime, which can be
predicted with the Re number. The Re number is the ratio
of inertial forces to viscous forces and can be calculated
with the equation
Re
rVDh
m
(1)
where r is the fluid density, V is the characteristic fluid

velocity, Dh is the hydraulic diameter of the microchannel,
and m is the fluid viscosity. Fully developed fluid flow in a
channel of circular cross-section is considered laminar if
the Re number is <2100. For Re numbers between 2100
and 2300, the flow is considered transitional: it shows
signs of both laminar and turbulent flow. A Re number
>2300 indicates turbulent flow.
Laminar flow is predictable in the sense that the trajectories of microscopic particles suspended in it can be
accurately predicted (Fig. 1a). Particles suspended in a
turbulent fluid flow behave chaotically and their position
as a function of time cannot be accurately predicted
MICROFLUIDICS
421
Figure 3. (a) At low velocities (low Re numbers), flow separation

will not occur in microchannels. (b) At higher velocities (larger Re
numbers), flow separation may become apparent.
Figure 1. (a) Particles suspended in a laminar flow within a

straight microchannel follow straight trajectories. (b) Particles
suspended in a turbulent flow within a straight microchannel
do not follow straight trajectories unless they are very close to
the wall.
(Fig. 1b). Fluid flow that has a Re number <1 is also known
as viscous flow, creeping flow, boundary-layer flow, or
Stokes flow.
Low Re number flow is best visualized by imagining
how honey (or any viscous substance) behaves when
poured or stirred. For example, water flowing in
microchannels will generally have a Re number <1. In
this case, water will behave like a very viscous liquid (i.e.,
like honey). An important point to make here is that the
properties of water do not change at the microscale; rather
the microscale dimensions involved make the water
appear more viscous than what we are accustomed to at
the macroscale. An excellent description of low Re number
environments has been given by Purcell (3). Very viscous
fluid flows have certain characteristics: the flow is reversible, mixing is difficult, and flow separation does not occur
(4).
Reversibility is the ability of a suspended particle in a
fluid to retrace its path if the flow is reversed. This is a
result of the minimal inertia (i.e., low Re number) present
in fluid flows at the microscale. Figure 2 shows the path a
suspended microscopic particle might take in forward and
reverse flow.
A second characteristic of microscale fluid flow is a lack

of flow separation. Flow separation is commonly observed
in the form of vortices, which are recirculating flows separate from the main flow. Because of the low Re number
environment, vortices usually will not form within microfluidic channels, as shown in Fig. 3a. Separation will only
occur in flows wherein inertial forces are significant relative to viscous forces (Re > 1). Figure 3b is a qualitative
sketch of flow separation in a cavity.
The third characteristic of microscale fluid flows is
inefficient mixing as a result of very low Re number flow,
and thus negligible inertia. Low inertia means that stirring is not effective and that mixing must be accomplished
by diffusion. At the macroscale, stirring minimizes the
diffusion distances between two or more liquids by distributing folds of the liquids throughout the volume.
Microscale methods of mixing have been developed that
take advantage of the unique properties of the scale and
improve the efficiency of mixing over simple diffusion;
examples include using three-dimensional (3D) channel
geometries, patterned channel surfaces, and pulsatile
flow (5).
Figure 4 shows two streams flowing down a microchannel side-by-side. Because of the low Re number environment the streams will only mix by diffusion. If the
flowrate is slow enough, the streams will eventually
become uniformly mixed across the whole microchannel
width.
The hydraulic diameter, Dh, of a microchannel is determined by its cross-sectional geometry and can be calculated
with the equation
Dh
Figure 2. (a) A suspended particle in laminar flow around an

obstacle in a microchannel. (b) If the flow is reversed, the particle
will retrace the same path.
4A
P
(2)
Figure 4. Two streams flowing in a microchannel will only mix by

diffusion. Note that the concentration across the width of each half
of the main microchannel is not constant as diffusional mixing
progresses.
422
MICROFLUIDICS
Table 1. Diffusion Coefficients for Biologically Important

Molecules in Watera
Molecule
Cl
O2
K
Na
Glucose
Lactose
Insulin
Hemoglobin
Urease
T, 8C
25
18
25
25
20
20
20
20
20
D, cm2 s1
5
2.03 10
2 105
1.96 105
1.33 105
6 106
4.3 106
1.5 106
6.3 107
3.4 107
Diffusion time, s
0.02
0.03
0.03
0.04
0.08
0.12
0.33
0.79
1.47
DP
All values are from Ref. 6. The time for each particle to diffuse 10 mm is
shown for comparison.
where A and P are the microchannel cross-sectional area

and wetted perimeter, respectively. The hydraulic diameter is often used to calculate important flow characteristics for noncircular microchannel cross-sections.
At microscale dimensions diffusion is an effective
mechanism for transporting molecules because of the
relatively short distances involved. Particles diffuse from
areas of high concentration to areas of low concentration
and will eventually diffuse to uniform concentration
throughout a given volume. The mean distance, d, a
particle travels in a time, t, can be predicted with the
equation
d2 2Dt
drop) will experience an internal pressure. This pressure is

called the YoungLaPlace pressure. Smaller fluid volumes
result in larger SAV ratios, thus increasing the internal
pressure. The pressure within a drop of liquid can be
calculated with the formula
(3)
where D is the diffusion coefficient of the particle.

Diffusion times are proportional to the square of
distance, which means that particles can diffuse across
microscale distances within a particular medium in a
matter of seconds. Table 1 lists representative
molecules of biological significance and their diffusion
coefficients.
The SAV ratios become very large at the microscale.
Typical SAV ratios for macroscale containers such as Petri
dishes or culture flasks are 10 cm1, while they are 800
cm1 for microfluidic channels. Increased SAV ratios allow
diffusion-limited processes, such as immunoassays to
become much more efficient at the microscale because of
the increased surface area available for binding. Large
SAV ratios also allow rapid heat radiation from microscale
fluid volumes and efficient gas exchange with the ambient
atmosphere and fluid in microchannels (assuming the
microchannel is made of a gas-permeable material).
Enhanced gas transport is a critical ingredient for cell
culture in microscale environments. One drawback of
large SAV ratios is that evaporation becomes a significant
problem.
The surface tension of a liquid becomes increasingly
important at very small dimensions. To visualize this,
think of a liquid surface as an elastic skin. If a slit were
made in that skin, a certain amount of force per unit length
would be required to hold the two sides of the slit together.
The amount of force required to hold the two sides together
is called the surface tension. Because the liquid surface is
under tension, liquid confined by the surface (e.g., a rain-
2g
R
(4)
where g is the liquid surface energy and R is the radius of

the drop. At microscale dimensions, significant pressures
can be created by surface tension. A common result of the
pressure difference of an air/liquid interface is the capillary effect: a pressure difference across the interface
propels liquid through a small diameter capillary or
microchannel.
The capillary effect also depends on the contact angle of
the microchannel surface. Hydrophobic surfaces (e.g., polymers) have contact angles >908 and hydrophilic surfaces
(e.g., glass) have contact angles <908. Microfluidic devices
with hydrophilic surfaces can be filled via capillary action.
The pressure difference at an airliquid interface within a
microchannel with square cross-sectional area can be calculated with the formula

cosuc cosuc
DP 2g
(5)
W
H
where W and H are the microchannel width and height,
respectively, and uc is the contact angle of the liquid on the
internal microchannel walls. Conversely, equation 5 gives
the pressure required to force water into a hydrophobic
microchannel of rectangular cross-section.
Microfluidic Modeling
Microscale fluid flow can be modeled from either a macroscopic or microscopic vantage point. Macroscopic modeling
treats the fluid as a well-mixed volume while the microscopic view looks at how particles suspended in the fluid
would behave under different flow conditions.
Macroscopic modeling, also called lumped modeling,
uses conservation of mass to predict microfluidic system
behavior. A pressure drop, DP, applied across a microchannel (or other conduit) with fluidic resistance Z, will induce a
volumetric flow rate Q:
DP QZ
(6)
All microchannels have a fluidic resistance associated

with them that depends on the geometry of the microchannel and the viscosity of the fluid. The fluidic resistance of a microchannel with a circular cross-section is
given by
Z
8mL
pR4
(7)
where m is the fluid viscosity, L is the microchannel

length, and R is the microchannel radius. The fluidic
resistance of a microchannel with a rectangular crosssection is given by
4mL a1
Z
f
(8)
b
ab3
MICROFLUIDICS
where f(a/b) is calculated with the formula

f
1
a 16 1024b X
tanh ma

5
b
3
ap n0 2n 15
(9)
When calculating the resistance of microchannels

with rectangular cross-section, m is the fluid viscosity, L
is the microchannel length, 2a and 2b are the microchannel width and height, respectively, and m is
calculated with
m
p2n 1
2b
423
Microscopic modeling is performed when precise modeling of fluid behavior is needed. For example, cells
attached to the wall of a microchannel might affect flow;
modeling at the microscopic level would reveal any perturbations of the flow caused by the cell. In contrast,
macroscopic modeling is performed when the behavior
of the entire microfluidic system is needed. For example,
fluid flow in many parallel microchannels might be
required. Macroscopic modeling would reveal the relative
flowrates through each microchannel and provide the
microchannel dimensions needed to guarantee equal flow
through each.
(10)
PUMPING FLUIDS
If the aspect ratio of the microchannel is very small (i.e.,

2b
2a) then the simplified formula
Z
3mL
4ab3
(11)
can be used. The general rule of thumb is that equation 11

should be used for microchannels with b/a <0.1.
The resistance of other geometries can be found elsewhere (7).
In predicting microfluidic system behavior, the analogies to Kirchhoffs laws are used. The sum of pressure drops
in a fluidic loop must be equal to zero; the total volumetric
flowrate entering a node must be equal to the total volumetric flowrate leaving a node.
In contrast to the macroscopic view, microscopic
modeling allows fluid behavior to be predicted. Specifically, the microscopic view allows the velocity profiles of a
fluid flow to be calculated. Velocity profiles are plots that
show the relative velocities of different portions of a fluid
within a microchannel. Figure 1a is an example of a
velocity profile.
The velocity of flow in a microchannel with circular
cross-section varies radially and can be predicted with
the formula

R2 DP
r2
vr
1 2
(12)
4mL
R
where m is the fluid viscosity, L is the microchannel length,
DP is the pressure drop, and R is the microchannel radius.
The velocity profile of flow in a microchannel with rectangular cross-section varies along the height and width axes
and can be predicted with the formula
!
1
DP
4X
n 1 cos my cosh mx
2
2
vx; y
1
b y
2mL
b n0
cosh ma
m3
(13)
where m is the fluid viscosity, L is the microchannel
length, DP is the pressure drop, m is calculated from
equation 10, and 2a and 2b are the microchannel width
and height, respectively.
Fluids are pumped through microfluidic channels by creating gradients; the two most common types being pressure
and electrical. Other types of gradients and their applications are discussed elsewhere (8).
Pressure gradients are the most common method used
to pump fluid. Pressure is applied to one end of a microchannel which causes the fluid to flow down the pressure
gradient. Common methods for creating a pressure gradient include pumps or gravity. Most methods for creating
pressure-driven flow use macroscale pumps attached to the
microfluidic device via tubing. Ideally, pumps should be
incorporated on-chip to realize the ultimate vision for LOC
devices. Many types of microfluidic pumps have been
demonstrated and they presently constitute an active area
of research (9).
Pressure-driven flow is attractive for use in microfluidics because it is easy to set up and model. Some drawbacks
for using pressure-driven flow are sensitivity to bubbles,
sensitivity to motion (via the tubing connecting pumps to
the microfluidic device), and parabolic flow profiles. Shear
stress is proportional to the pressure drop across a microchannel, which should be taken into account when manipulating cells.
The other common way to pump fluids is to use electrical
gradients. This method of pumping fluid is only practical at
the microscale level because of the large electric fields and
SAV ratios required. Pumping via electric fields is called
electrokinetic flow and is based on two phenomena: electrophoresis and electroosmosis. Electrophoresis operates
on the principle that charged particles in an electric field
will feel a force proportional to the field strength and their
charge. The particles will move through the electric field
toward the pole of opposite charge. Larger particles move
slower than smaller particles because of the drag produced
by moving through a fluid. Figure 5a shows an example of
electrophoretic flow.
A charged particle in an electric field of strength E will
travel with a velocity equal to
v mep E
(14)
where mep is the electrophoretic mobility. Electrophoretic

velocities are typically much smaller than the velocities
caused by electroosmosis.
Electroosmosis will only function in the presence of
an electric double layer at the surface of the microchannel.
424
MICROFLUIDICS
Figure 5. (a) Electrophoresis. Charged

particles will move toward oppositely
charged poles in an electric field. (b)
Electroosmosis. Charges lining a
microchannel surface will move
with an applied electric field, thus
inducing bulk flow via momentum
transfer within the liquid. (c) An
electric double layer forms at
charged microchannel surfaces; the
layer thickness is called the Debye
length.
An electric double layer forms at a charged surface

when oppositely charged particles from an electrically
neutral liquid gather at the surface. The thickness of
the electric double layer is known as the Debye
length; the concentration of charged particles at a
surface falls off rapidly as a function of distance and is
shown in Fig. 5c. When an electric field is applied across
the length of the microchannel, the ions gathered at the
microchannel surface begin to slide toward the oppositely
charged pole, as shown in Fig. 5b. As the ions slide, they
drag their neighbors within the bulk liquid, toward the
middle of the microchannel. The friction between subsequent sliding layers of ions causes the bulk fluid to begin
moving.
If the Debye length is much less than the characteristic
dimensions of the microchannel, then the bulk fluid velocity can be predicted with the equation
v meo E
very rapid response times, since the electrodes are integrated on chip, and are generally insensitive to
movement off chip. Drawbacks to electrokinetic flow
include fouling of the electrodes, which reduces electric
field strength and therefore flowrate. Also, protein adsorption to microchannel surfaces affects the Debye layer, and
in turn flow. Unintended side effects from electric fields on
biological cells within the microfluidic device may also
exist. Figure 6 shows the difference between the parabolic
flow profile of pressure-driven flow and the blunt profile of
electrokinetic flow.
Other methods of fluid flow based on surface tension,
heat, and evaporation, have also been demonstrated. However, these methods have yet to be widely adopted and it is
(15)
where E is the electric field strength and meo is calculated

with
meo
ez
4pm
(16)
where e is the dielectric constant of the fluid, z is the zeta

potential of the surface, and m is the fluid viscosity.
Electrokinetic flow is attractive because it only requires
the integration of electrodes in a microfluidic device, which
is straightforward by microfabrication standards. Electrokinetic flow is therefore amenable to interfacing with
electronic control circuitry. Electrokinetic flow also
results in a blunt flow profile, which reduces the distortion
of transported samples. Lastly, electrokinetic flow has
Figure 6. (a) Pressure-driven flow is parabolic; the middle of the

stream moves faster than regions near the wall. (b) Electrokinetic
flow is blunt; all parts of the stream move at equal velocity. Note
that residual pressures can cause the profile to become slightly
parabolic in the direction of the negative pressure gradient.
MICROFLUIDICS
425
Figure 7. (a) Material is etched

from a substrate and an enclosed
structure is formed by bonding the
etched substrate to a glass lid. (b)
Walls of microchannels are built on
top of a substrate and a glass lid is
placed on top of the photoresist to
form an enclosed structure.
unclear if they will prove more attractive than pressuredriven or electrokinetic flow.
FABRICATION
Fabricating microfluidic channels in traditional microfabrication materials, such as silicon and glass, can be
achieved in two different ways. In the first, material is
selectively removed, or etched, from a bulk substrate. The
etched substrate is then bonded to another material (e.g.,
glass or silicon), which may have access holes or other
features embedded in it. The result is an enclosed channel
structure, as shown in Fig. 7a. The second method is to
selectively add material to a substrate, and then bond
another substrate to it. This method will also form enclosed
channel structures as shown in Fig. 7b.
Photolithography is a fundamental part of all
microfabrication (Fig. 8). Light is used to project patterns
onto a photosensitive chemical, called a photoresist. The
photoresist can be either positive or negative. Light
chemically alters positive photoresist and makes it soluble
in a developer. Negative photoresist is cross-linked by
light, which makes it insoluble in developer. The patterned photoresist can be used as an etch mask for substrates, producing microchannels like those shown in
Fig. 7a. Patterned photoresist can also be used in subsequent steps to direct the patterning of other materials that
cannot be directly patterned with light. In-depth treatments of microfabrication techniques can be found elsewhere (10,11).
Polymers have recently become popular alternatives to
traditional (e.g., silicon or glass) microfabrication materials. Polymers can be used to make microchannels by using
the same methods mentioned previously for silicon and
glass. Polymers also have the advantage that they can be
molded that makes them a cheaper alternative (relative to
silicon or glass) for mass production.
Polymer microfluidic devices can be created by molding,
hot embossing, injection molding, photopolymerization,
and laser ablation or laser cutting. An attractive method
for fabricating polymer microfluidic devices is to use a
process known as micromolding (12). In this process, photolithography is used to make a pattern of the microchannels
(called a master). The photoresist provides a positive relief
from which a polymer mold can be cast. The polymer is
poured over the master and allowed to cure. The polymer
mold is then peeled from the master and either placed on a
substrate or incorporated into a multilayer device. The two
advantages of this microfabrication method are (1) no
special microfabrication equipment is required, and (2)
many inexpensive copies of a microfluidic device can be
rapidly manufactured.
Drawbacks to using polymers include leaching of material into microfluidic channels, solvent incompatibility,
Figure 8. Photolithography requires an ultraviolet (UV) light

source, a mask, and a photosensitive layer of material (i.e.,
photoresist). The photoresist is
patterned with the UV light via a
mask.
426
MICROFLUIDICS
Figure 9. (a) The Bioanalyzer uses

microfluidic chips etched from glass.
(Images courtesy of Agilent Technologies). (b) The etched glass chips are
encased in a plastic assembly that
facilitates handling and limits
contamination. Images courtesy of
Agilent Technologies. (c) Samples
are loaded onto the chip and the
chip is loaded into the Bioanalyzer.
(Images
courtesy
of
Agilent
Technologies). (d) The Bioanalyzer
then performs an analysis on the
samples. (Images courtesy of
Agilent Technologies.)
and the ability of some substances to diffuse into the

polymer. Also, surface treatments are occasionally
required to make polymers compatible with electrokinetic
flow; silicon or glass have an inherent surface
charge that allows them to be used in electrokinetic flow
applications.
BIOMEDICAL APPLICATIONS OF MICROFLUIDICS

Microfluidic concepts have already been incorporated in a
variety of biomedical devices (13). One example of a microfluidic device now found in many biomedical labs is
Agilents Bioanalyzer. The Bioanalyzer system uses disposable chips etched in glass. The samples to be separated
and reagents for the separation are loaded onto the chip via
reservoirs. The chip is then placed in a reader, where
electrokinetic flow is used to manipulate the samples in
the reservoirs (Fig. 9).
Microfluidic capillary electrophoresis systems are
becoming commonplace in laboratories. By using
very small volumes for separation, joule heating from
electrophoresis is rapidly radiated away from the gel.
Efficient heat radiation allows larger voltages to be used

which translates into faster separations. The shorter
separation distances used also contribute to reduced analysis times. Microfluidic capillary electrophoresis systems
allow DNA to be rapidly analyzed, and have highly reproducible results since the entire process is automated.
Lastly, contamination is minimized because the devices
are disposable.
Because of their small size, another attractive
aspect of capillary electrophoresis (CE) systems is that
they can be incorporated into LOC devices and made
part of a complete system. An example that has been
demonstrated in several research labs is a system that
takes cells as inputs, lyses them, performs all necessary
preprocessing, DNA amplification, and so on, and then
performs the DNA separations, all with no human intervention (14).
Microfluidics are also being used in clinical devices;
devices for hematology and disposable assays for pointof-care diagnostics are among those now being researched
and brought to market. A handheld point-of-care device
made by i-STAT is an example of a clinical microfluidic
device (Fig. 10). The handheld device quantifies analytes in
MICROFLUIDICS
(a)
(b)
427
idge
Cartr
Cartridge label
Sample entry
well gasket
Fluid channel
Cartridge cover
Figure 10. (a) The handheld pointof-care device manufactured by iSTAT performs analyses on blood
samples contained in a disposable
cartridge. Reprinted with permission from ACS (from Ref 15).
Copyright 1998 American Chemical Sa. (b) Microfluidic cartridges
are loaded with a sample and then
plugged into the i-STAT handheld
device. Image courtesy of Abbott
Point-of-Care. The cartridge contains all necessary microfluidic
control and sensing components
that are then actuated by the
handheld device.
Sample entry well
Tape gasket
Biosensor chips
Calibrant pouch
Puncturing barb
Cartridge base
Air bladder
blood samples that have been deposited on a disposable

chip. The general procedure for operation is given
below (15).
A patients blood sample is deposited in a well on the
disposable chip and then the well gasket is snapped shut
(Fig. 10a). The microfluidic chip is then inserted into a
handheld reader that performs an automated analysis of
the blood sample, as shown in Fig. 10b. On-chip biosensors
are automatically calibrated and checked for accuracy
with an on-chip packet of calibrated solution. Once
their accuracy has been determined, the calibration solution is flowed to the waste compartment. The blood
sample is then flowed over the biosensors and the concentrations of different analytes are displayed on the handheld device screen within a few minutes. Diaphragm
pumps are used to move the fluid. A variety of chips are
available for different assays, including electrolytes and
blood gases.
CONCLUSION
Microfluidics is the study and application of fluids at
the microscale. Techniques used by the microelectronics
industry have been adapted to facilitate the creation of
micron-size channels capable of carrying fluid. The
physical behavior of fluid at the microscale differs from
behavior observed at the macroscale in everyday experience. The miniaturization of fluid handling has allowed
LOC devices to be created in which all of the procedures of a
traditional chemistry or biology lab are performed automatically in a single microfabricated chip. Lab-on-a-chip
devices will allow new clinical and research tools to be
developed.
BIBLIOGRAPHY
1. Brody JP, et al. Biotechnology at low Reynolds numbers.
Biophys J 1996;71(6):34303441.
2. Karniadakis GE, Beskok A. Micro Flows. 2nd ed. New York:
Springer-Verlag; 2001. p 360.
3. Purcell EM. Life at low Reynolds number. Am J Phys
1977;45(1):311.
4. Meldrum DR, Holl MR. Tech.Sight. Microfluidics. Microscale
bioanalytical systems. Science 2002;297(5584):11971198.
5. Nguyen NT, Wu ZG. Micromixersa review. J Micromech
Microeng 2005;15(2):R1R16.
6. Stein WD. Channels, Carriers, and Pumps: An Introduction
to Membrane Transport. San Diego: Academic; 1990.
7. Shah RK, London AL. Laminar Flow Forced Convection in
Ducts. New York: Academic; 1978.
8. Stone HA, Stroock AD, Ajdari A. Engineering flows in small
devices: Microfluidics toward a lab-on-a-chip. Ann Rev Fluid
Mech 2004;36:381411.
9. Laser DJ, Santiago JG. A review of micropumps. J Micromech
Microeng 2004;14(6):R35R64.
10. Kovacs G. Micromachined Transducers Sourcebook. Boston:
WCB McGraw-Hill; 1998.
11. Madou M. Fundamentals of Microfabrication. 2nd ed Boca
Raton(FL): CRC Press; 2002.
12. McDonald J, et al. Fabrication of microfluidic systems in
poly(dimethylsiloxane). Electrophoresis 2000;21(1):2740.
13. Beebe DJ, Mensing GA, Walker GM. Physics and applications of microfluidics in biology. Ann Rev Biomed Eng
2002;4:261286.
14. Waters L, et al. Microchip device for cell lysis, multiplex PCR,
amplification, and electrophoretic sizing. Anal Chem 1998;
70(1):158162.
15. Lauks IR. Microfabricated biosensors and microanalytical
systems for blood analysis. Acc Chem Res 1998;31(5):317324.
See also DRUG DELIVERY SYSTEMS; NANOPARTICLES
428

JI YOON KANG
Korea Institute of Science and
Technology
Seoul, Korea
INTRODUCTION
Generally, micropower is the local generation of electricity
by small-scale generators, which locates the end point.
As the recent development of the microelectromechanical
system (MEMS), as well as CMOS electronics technology,
has been reducing the size and cost of biomedical devices,
the research of micropower became important for implantable biomedical devices since they require internal selfsustained power sources.
As for biomedical devices, micropower is an internal or
external power source to supply energy for active devices,
which replaces an organs function or treats diseases. The
examples of active implantable devices that consume energy
are cardiac pacemakers, cardiac defibrillators, muscle stimulators, neurological stimulators, cochlear implants, and
drug pumps (1). Hence, in this article the term micropower
describes rather tiny power supplying devices for miniaturized sensors, actuators, and electric devices, whose size is
> or < 1 cm3.
The low power electrical actuator, such as a pacemaker
or neuronal stimulator, requires tens of microwatts intermittently and their power source is usually a lithium iodine
battery that lasts from 58 years. Usually, the stand-by
current of a pacemaker is 1 mA in waiting mode and its
pulse current is 6 mA. One example of a specification for
a pacemaker pulse is in the range of 25 mJ (11 mA at 2.2 V
with a 1 ms discharge) and the capacity of the battery is 2
Ah at typical rating (2). The volume of a pacemaker is
20 mL and the volume occupied by the battery is about
one-half of the total volume, 10 mL. Hence, the energy
density (energy/volume) and reliability are important factors in the lifetime of the device.
An internal battery that is hermetically sealed in these
devices can operate them with low power consumption;
however, other implantable devices have radically different power requirements. Implantable cardioverter defibrillators demand the energy of 1540 J providing six orders of
magnitude larger than that of a pacemaker even though
the pulses are less frequent. The current from a lithium
silver vanadium battery is charged in an internal capacitor
and the pulse of 12 A of current is fired. Electromechanical
actuator like drug pumps demand more current than a
lithium ion battery can deliver since it needs to overcome
the high pressure in the chamber. The examples of drug
pumps are insulin pumps, pain reliever, and an cerebrospinal fluid pump. A high current implantable battery
should have low source impedance, such as lithium thionyl
chloride, lithium carbon monofluoride, or lithium silver
vanadium oxide.
Other future application are in wireless sensors, including physiological, chemical, or physical sensors embedded
in an encapsulated environment. Miniaturized sensor
consumes < 100 mW and a radio frequency (rf) transmitter

consumed 10 mW intermittently. Since most of the power
is used for communication, some research groups are
developing several low power wireless transmission protocols (3,4). Hence, less power will be necessary for a sensor
transmitter as technology evolves.
Some groups investigated more efficient and reliable
batteries. To enhance their efficiency and lifetime, potential alternatives of the conventional batteries studied, such
as a microfabricated battery, microfabricated fuel cell, and
biofuel cell. Microelectrical system technology reduces the
size of the primary battery and microfluidic galvanic cell
(5), water activated microbatteries (6), and Li-ion microbatteries were demonstrated. The fuel cell attracts much
attention since it has a high efficiency, high power, and low
pollution rate. Research on fuel cells focus on high power
applications, such as the automobile and portable electronics,
like laptop computers and cellular phones (7). Recently,
micromachining technologies employed as a method to fabricate miniature fuel cells (811) and their size became
smaller than a button cell battery (12) with high power.
Enzyme-based glucose/O2 biofuel cells were reported by
several groups (13) and a miniaturized all (14) was reported
that is < 1 mm3, with although a power of 4.3 mW. Since
glucose is available in all tissues and organs, it is advantageous in implanted medical sensortransmitters.
Although a primary battery as well as a rechargeable
battery is an important tool that supplies reliable power to
implant devices, the continuous power of the battery
decreases with time, and after 5 years they will not supply
enough power (15). Power delivery with an rf transmission
can extend the lifetime and continuously deliver high power.
In the case of a cochlear implant, an external device provides
power and data through electromagnetic field coupling;
however, it needs accurate positioning of the external device
and may cause rf interference and heating of the tissue.
Therefore, many research groups are paying attention to
microfabricated power scavenging devices as an auxiliary
power source to recharge the battery with no external power.
Ambient energy sources are body heat or movement of the
human body. Piezoelectric material (1618), capacitance
change (1924), and inductive coil (2527) convert vibration
or human motion into electrical energy. The generation by
high frequency vibration is not suitable for implant devices
since vibration of the human body is in the range of tens of
hertz. Hence, energy conversion using vibration of low frequency can be integrated with implant devices.
Thermoelectric generators that convert temperature differences of the human body or combustion engine to electricity were reported (28,29). Another conversion method is
the thermophotovoltaic power generator (30,31), which combines the combustion engine and solar cell. However, integrating a power generation device into an implant device
has some limitations due to biocompatibility. Power generation with a high temperature like the combustion engine or
thermovoltaic metoid, cannot be implemented in the inside
of the human body due to the heating of tissues. Thermoelectric generation using body heat is promising for the
implantable device. However, this article includes the
review on the other portable power sources like the micro-
combustion engine, because those are also useful for a

portable diagnosis system. The recent development of microfluidics and miniaturized biosensors enables point-of-care
testing devices to be on the market in the near future. A
microheat engine is highly efficient in energy conversion
and will be useful as a portable medical equipment.
A good review article on micropower for wireless sensor
networks was reported (20,32). It lists the candidates of
portable power sources and compares the energy density for
the battery and power density for a power generator. This
article reviews the existing and potential micropower sources
in view of medical applications from tiny sensors embedded in
the human body to portable medical electronic devices.
MICROBATTERY
For a long time, the battery was a major energy source in
portable electronic devices and it has evolved from Zn/
MnO2 to the Zn/air cell since 1900. Electrochemical power
sources were developed in response to the needs of the
flashlight, automotive starter, mobile electronics, and laptop computers. These days, there is a tremendous need for
portable electronics demanding smaller, lighter, and longer
lived batteries. Hence, many researches are on the way to
making microfuel cell or microfabricated batteries using
various kinds of electrochemical power. This section will
briefly review microbatteries including fuel cells, biofuel
cells, and micromachined batteries.
Microfuel Cell
Although the energy density of conventional batteries has
been increasing from 500 Wh L1 for the Ni/Cd battery
to 1500 Wh L1 for the Li/CCoO2 battery, there is a
large jump for the air-cathode fuel cells of 4500 Wh L1
using hydrogen, hydrocarbon, and metals (7). The proton
exchange membrane fuel cell (PEMFC) depicted in Fig. 1 is
one of the promising techniques for fuel cell, which was
Power
Water
H2O
H2
H+
Hydrogen
O2
Oxygen
Anode
Electrolyte
Cathode
Figure 1. Schematic of a PEM fuel cell.
429
implemented in miniaturization. Power sources for the

automobile or the portable electronics of a huge market
have been a main concern of fuel cell research because of
the advantage of high efficiency and easy rechargeability.
However, these days, in response to the demand of low
power application, many miniaturized fuel cells are under
study. Miniature fuel cells with a series path in a flipflop
configuration was fabricated in a planar array and a fourcell prototype was produced, 40 mW cm1 (33). Yu (11)
added microfluidic channels using anisotropic wet etching
of silicon to the flipflop configuration and measured a peak
power density of 190 mW cm2. They also reported that
the flipflop fuel cell was constructed on printed circuit
board (PCB) and that they achieved the area power density
of >700 mW cm2. Wainright (12) fabricated on-board
hydrogen storage with multiple coplanar fuel cells in series
on ceramic substrates. They stored hydrogen in the form of
the stabilized aqueous solutions of sodium borohydride
(NaBH4) or a metal hydride material, such as LaAl0.3Ni4.7.
The energy density of microfabricated fuel cells did not
exceed that of a Li/MnO2 coin cell, but it had the advantage
of higher power and compatibility with other microelectronic, microelectromechanical, or microfluidic devices. A
polymer microfluidic channel was applied to a fuel cell
using PDMS (10) and poly (methyl methacrylate) (PMMA)
(8). They achieved a comparable area power density with a
silicon based one. Whitesides and co-workers (9) also
reported a membraneless vanadium redox fuel cell using
a laminar flow property in a microfluidic channel with
200 mW cm2. As for the commercialization, MTI microfuel cells and Manhattan Scientifics are making portable
fuel cells using direct methanol fuel cell (DMFC) and Medis
technology announced direct liquid fuel cell (DLFC) for
handheld devices. Miniaturized fuel cells are promising for
implantable medical devices with a high power requirement and it has the advantage of a longer lifetime and less
charging time than a conventional battery.
However, PEM fuel cells do not fit well with the implantable microdevices with high power and a long life application because the refill of a hydrogen fuel cell is not easy
when it is sealed in the human body. As an alternative fuel
cell, the biofuel cell is a strong candidate for a low power
embedded device like a microbiosensor. Although a biofuel
cell is not capable of high power, the easy availability of fuel
(glucose) gives it a long operation time. Enzyme-based
glucose/O2 biofuel cells were studied since 1980s (13)
and Heller and co-worker (34) demonstrated the feasibility
of a membraneless miniature biofuel cell as an implanted
micropower source. Its chemical reaction is described in
equations 1 and 2.
glucose ! glucolactone 2H 2e
(1)
O2 4H 3e ! 2H2 O
(2)
Some fuel cell components, such as case, membrane, ion

conductive electrolyte, and plumbing was removed in the
biofuel cell and its size became <1 mm2. The power of the
biofuel is 4.3 mW with 0.52 V and it is suitable for an
implanted devices because of tiny volume and abundant
glucose inside of human body. Moor et al. (35) developed a
microfluidic chip based ethanoloxygen biofuel cell, which
430
produced 18 mW with 0.34 V and is applicable to integrate

the biofuel cell and the microfluidic chip.
Micromachined Battery
A microfabricated battery usually refers to a thin-film
battery, however, recently some research groups are studying MEMS-based microbatteries. Integration of microbatteries with CMOS electronics or an MEMS device is easy
and can be fabricated on a chip with a device. Since the
microbatteries main concern is power output due to limitation of surface rather than the capacity, most research
activities are focused on increasing power to out perform
maintaining capacity.
As for thin-film batteries, Bates et al. (36) at Oak Ridge
National Laboratory reported a thin-film secondary battery, which was made up of lithium and lithium ion. The
thickness is tens of mm and the area is in the cm2 range. Its
continuous current output is 1 mA cm2. Pique et al. (37)
constructed primary ZnAg2O and secondary Li ion microbatteries in plane using laser direct-write with a capacity
of 100 mAh cm2.
Other than classical thin-film batteries, several micromachined MEMS batteries were developed. They try to
integrate power sources with microelectronic circuits and
microsensors. Prof. Lin at UC Berkeley proposed a wateractivated battery with 1.86 mWh in the area of 12 12 mm
for lab-on-a-chip application (6) that overcomes the corrosiveness of the micromachined batteries with sulfuric acid
and hydrogen peroxide (38). Andres (5) devised a pump
integrated with a micropower source, in which microfluidic
galvanic cells supplied power to heat up the two-phase fluid
for pumping. The capacity of the micromachined battery is
lower than that of a thin-film battery yet, its application
is restricted to the integrated power for a micromachined
implantable device.
MICROPOWER GENERATOR
A micropower generator scavenges energy from devices.
The energy sources are mechanical (vibration and human
body movement), thermal (temperature difference), and
solar energy. Thermoelectric devices convert the temperature difference to electricity and the photovoltaic cell
collects solar energy. A MEMS-based power generator
scavenges energy from the vibration in the mechanical
structure or human body movements.
Thermoelectric Generator
Thermoelectric generation using the Seebeck effect was
widely studied. This effect was discovered in 1821 by the
physicist, Jonn Seebeck. It is the same phenomenon with a
thermocouple where the temperature difference produces
electricity or work. Although a thermoelectric power generator is an old technique and is commercially available in
various sizes in the market, recent research focuses on
making miniaturized low power thermoelectric microgenerators using microfabrication. Cost-effective fabrication technology was developed using electroplated
structures with an epoxy film (28) and nanowire arrays
by electrochemical deposition was implemented to improve
thermoelectrical properties (29). Reportedly, several companies announced a thermoelectric generator using body
heat. Applied Digital Solutions (39) announced a thermoelectric generator called ThermoLife, which produced a
power of 49 mW from a temperature difference of 5 8C in
0.5 cm2. Biophan technologies (40) also announced a
biothermal power source as shown in Fig. 2 for implantable
devices like a pacemaker and defibrillator (41). They aim to
produce 100 mW at 3 V with 1 8C temperature difference. A
different scheme converting thermal energy to electricity is
piezoelectric generator actuated by thermal expansion of
two-phase working fluid (42). Although they produced up to
56 mW at its resonance frequency of 370 Hz, a practical
application needs the careful design of a heat-transfer
mechanism because the temperature is required to
oscillate at a resonance frequency of structure. A thermoelectric generator has a well-established technology and
its operation is relatively stable; however, note that the
temperature difference inside the human body is < 1 8C.
The temperature gradient is maximum at the skin surface
and will limit the location of the thermoelectric power
generator.
p-type Semiconductors pallets

Ceramic
substrate
Heat absorbed
(Cold side)
n-type Semiconductors pallets
Conductor
Positive(+)
Heat rejected
(Hot side)
Negative()
Figure 2. Principles of thermoelectric power source

(Biophan).
431
Continuous Power / cm3 vs. Life Several Energy Source

1000
Solar
Lithium
Alkaline
microWatts
100
Vibrations
10
Zinc air
Lithium rechargeable
1 NiMH
0
0
0.5
1.5
2.5
3.5
4.5
Years
Power Generation with Ambient Vibration

Conversion of a mechanical vibration to electric power has
been studied from the mid 1990s (43) using MEMS, while
thermal and solar energy were exploited to generate electricity long ago. They changed vibration to electrical
energy using piezoelectric, electromagnetic, and electrostatic generations.
The frequency of vibration in an ambient environment
ranges from 60 to 400 Hz and for the microwave oven the
acceleration was 2.25 m s2 at a resonance frequency of
120 Hz (15). Shad (21) suggested a graph comparing the
power density of power scavenging and batteries as a
function of time (Fig. 3). Power density in the vibration
of machining center or microwave oven becomes larger
than conventional batteries after 3 or 4 years, which means
the waste vibration energy is not negligible. Williams and
Yates (43) presented a general model in equation 3 for the
power of external vibration as depicted in Fig. 4, when a
mass is moved at a resonant frequency of vn.
P
zt mYo2 v3n
4zt zo 2
(3)
where m is mass, and Yo is the maximum extent that the

mass can move, zt and zo denote a damping factor both
Figure 3. Comparison of power density

from vibrations, solar and batteries (21).
in the transducer structure and in the environment

(e.g., air).
Electromagnet Conversion. Motion between the inductor and the permanent magnet induces an electromagnetic
current in the inductor coil, as shown in Fig. 5 (25). The
induced voltage, V, in the coil is given by equation 4.
vn
2z
V NBl
(4)
where N is the number of turns in the coil, B is the

strength of the magnetic field, l is the length of coil,
and z is the displacement of the magnet in the coil. This
type of generator was fabricated with laser micromachining by Ching et al. (27) in 1 cm3 volume and generated a
4.4 V peak-to-peak with a maximum rms power of 830 mW.
Glynne-Jones et al. (44) at the University of Southampton, derived 157 mW on average new car engine. Perpetuum Ltd., a spin-off company from the University of
Southampton, produced an electromechanical microgenerator. That generated up to 4 mW and its operation
frequency was 30350 Hz. The vibration amplitude was
200 mm with 60110 Hz and demonstrated a wireless temperature sensor transmitter system. This result showed
electromagnetic conversion is feasible in low frequency
vibration and is promising if it is compatible with silicon
micromachining.
k
k
z
m
+
z
m
Permanent
magnet
y
y
Figure 4. Schematic analysis for mechanical movement of a

power generator with external vibration.
Figure 5. Schematic of an electromagnetic conversion device.
432
Anchor
SW1
SW2
PZT layer
Vin
Silicon layer
Cv
Cpar
Cstar
Mass
Figure 6. Structure of a piezoelectric cantilever for power

generation.
Piezoelectric Conversion. The piezoelectric effect states

that the deformation in the material produces an electrical
charge due to the separation of charge within crystal
structures. The most widely used piezoelectric material
is PZT (lead zirconate titanate) in ceramic materials and
PVDF [poly (vinylidene fluoride)] in polymers. Several
groups are studying the piezoelectric cantilever with
seismic mass (Fig. 6). The constitutive equations of
piezoelectric materials are expressed in equation 5.
s Yd d31 E
D d31 s eE
(5)
where s is the mechanical stress, d is the mechanical

strain, Y is Youngs modulus, d31 is the piezoelectric strain
coefficient, D is the charge density, E is the electric field,
and e is the dielectric constant of the piezoelectric material. The piezoelectric coefficient links the mechanical
stressstrain to the electrical charge equation. If the
circuit is open (D 0), the voltage across the piezoelectric
layer is described in equation 6.
V
d31 s
tpiezo
e
(6)
where tpiezo is the thickness of the piezoelectric layer. The

charge collected on the electrode is integrated on the area
of the surface with no load condition as in equation 7
Z
Z
Q D dA d31 s dA
(7)
When the impedance in the load circuit is pure resistance,
the time-averaged power can be derived in Ref. 17 with
the geometry of a cantilever. White and co-workers (16)
presented a thick-film PZT generator, and the maximum
power is 2 mW. According to the analysis of Lu et al. (17),
a 5 mm long PZT cantilever can generate >100 mW with
the amplitude of >20 mm at 3 kHz resonance. Roundy
(15) demonstrated a piezoelectric converter of 1 cm3 in
volume, that is 1.75 cm in length. It generated 200 mW and
is driven with vibrations of 2.25 m s2 at 120 Hz. A
microfabricated PZT cantilever generator driven by a
bubble was studied by Kang et al. (44), and a few picowatt
was generated with one tiny cantilever at 30 Hz and tens
of mW is expected on a 1 cm2 surface (46). If the design,
material, and fabrication are optimized, piezoelectric
powergeneration will produce hundreds of microwatts
with a volume of 1 cm3.
Figure 7. Circuit representation for an electrostatic converter

(15).
Electrostatic Conversion. The electrical energy in a

capacitor is given in equation 8.
1
1
1 Q2
E QV Cv V 2
2
2
2 Cv
(8)
When the charge, Q, is constant, if the variable capacitance, Cv, is decreased the total energy E in the capacitor
will increase. The MEMS structure can change the capacitance Cv with an external vibration, and stored energy in
the capacitor transfers to energy storage. In the beginning,
the external power source Vin initiates the charging process as in Fig. 7 (15). When Cv is maximum, SW1 is closed
and the variable capacitance Cv is charged. While vibration
changes capacitance Cv, all switched are open. When Cv
reaches a minimum, SW2 is turned on and the energy in Cv
is transferred to a storage capacitance Cstor. The disadvantage of electrostatic conversion is that it needs an external
voltage source and switching circuit. The voltage across the
storing capacitance is given in equation 9.
1
Estor Cmax Cmin Vmax Vin
2
Ref: 47
(9)
where Vmax is the maximum voltage across the capacitor

Cv. Switching the circuit is realized using a diode and field
effect transistor (FET) switch. Meninger et al. (47) made a
comb-type variable capacitance with an in-plane overlap
type with a 7 mm gap and a 500 mm depth using the 0.6 mm
CMOS process. They produced a power of 8 mW with an
ultralow power delay locked loop (DLL)-based system. Miao
et al. (23) reported an out-of-plane variable capacitor with a
gap closing type that varies from 100 pF to 1 pF. A periodic
voltage output of 2.3 kV (10 Hz) was generated when the
charging voltage was 26 V, which implies that a power of 24
mW (2.4 mJ cycle1) can be produced. Mitchenson et al.
analyzed architectures for vibration-driven micropower
generators (26) and they fabricated a prototype of an electrostatic power generator producing 250 V cycle1 that
corresponds to 0.3 mJ cycle1 (22). Other studies demonstrated polymer capacitor (24) and a liquid rotor power
generator with a variable permittivity producing 10 mW
(19). Recent developments in electrostatic generators
433
analytical equipment. Another scheme of a heat engine is

thermophotovoltaic power generation (30,31). They converted the heat radiation in a SiC microcombustor to
electric energy using photovoltaic cells, which is <1 cm2
and produced a power of 1.02 W with 2.28 V.
CONCLUSION
Figure 8. SEM of microturbine of MIT (52).
demonstrated that it is feasible and manufacturable with

MEMS and that it has the advantage of a low frequency
application like human body movement. However, the generated voltage is very high and should be managed for the
implant application. Since it is compatible with the CMOS
process and the variable capacitor is a well-established
MEMS device, it is very promising as an integrated power
generator with a sensor and transmitter.
Microheat Engine
A microheat engine is hard to be implanted in the human
body, but it is promising as an external power source for
portable medical equipment. A tiny internal combustion
engine, made by precise machining, such as electrical
discharge machining (EDM) or MEMS, may have a much
higher density than a primary battery since the energy
density of fossil fuel is 45 MJ kg1, while that of Li ion
batteries are at most 0.5 MJ kg1 (48). Microturbine by
EDM (48), Microrotors by deep reactive ion etching (DRIE)
(49), heat engine (42) and reciprocating devices (50) was
reported for electric power generation.
Several groups are working on a microheat engine, since
fossil fuel offers a much higher energy density. Since the
generated power is expected to generate 1020 W, it is
suitable for high power application. The Stirling engine
(51), the reciprocal combustion engine (50), the Wankel
motors, and gas turbines are reported. One of the first
microengine projects was started at MIT (52) and the
microfabricated turbine with a 4.2 mm diameter was illustrated in Fig. 8. Massachusetts Institute of Technology
has been working on making microheat engines with a
turbo charger and Georgia tech is collaborating with MIT
on a magnetic generator (53). Allen and co-workers (54) at
Georgia Institute of Technology generated a direct current
(dc) electric power of 1.1 W with microfabricated windings
at 120,000 rpm, although it was not integrated with a heat
engine. Peirs et al. (48) made a microturbine by EDM and
that tested to speeds of 160,000 rpm and produced a
mechanical power of 28 W and an electrical power of
16 W. A miniaturized heat engine is still in the initial
stage and no demonstration of power generation using
a heat engine was reported. The heat engine would be
very useful for high power applications such as portable
This article reviews micropower devices for medical

implantable devices and portable medical device. Since
the micropower devices have their own characteristics,
there is no winner among them. When a selecting a micropower system for a specific application, one should consider
the energy capacity, power, volume, voltage, and compatibility of fabrication with microelectronic device.
Currently, primary or secondary batteries with external
power transmission are a main power storage for a low
power implanted device. As the application of implant
devices is diversified and requires a high power output
and longer lifetime, new battery like fuel cell, or biofuel will
replace conventional battery system in the future. Furthermore, when power transmission through the skin is impossible due to the attenuation of transmission, integrated
power generation device will be an alternative.
Most micropower generators are still in their infancy
and they need much more study to be implemented in
implant device. Research results on micropower generators
showed only the feasibility of concept and their power is
much less than the requirement. Hybrid micropower supplies (55) or integrated power system would be strong
candidates for long battery life applications. The promising application of active implant device will be glucose
sensor and the artificial pancreas for the treatment of
diabetes and the ubiquitous bio- or environmental sensor
network. Since there is strong demand in the market, it
is believed that micropower system will be available in
near future.
BIBLIOGRAPHY
1. Soykan O. Power sources for implantable medical devices.
Business briefing: Medical device manufacturing and technology. 2002, p 7679.
2. Mallela VS, Ilankumaran V, Rao NS. Trends in Cardiac
Pacemaker Batteries. Indian Pacing Electrophysiol J
2004;4: 201212.
3. Zhong LC, Shah R, Guo C, Rabaey J. An Ultra-Low Power
and Distributed Access Protocol for Broadband Wireless
Sensor Networks. Presented at IEEE Broadband Wireless
Summit, Las Vegas (NV); 2001.
4. David Culler DE, Srivastava M. Overview of Sensor Networks.
IEEE Comput Special Issue Sensor Networks 2004; 4149.
5. Cardenas-Valencia A, et al. A microfluidic galvanic cell as an
on-chip power source. Sensors Actuator B 2003;95:406413.
6. Sammoura F, Lee Kb, Lin L. Water-activated disposable and
long shelf life microbatteries. Sensors Actuators A 2004;111:
7986.
7. Dyer CK. Fuel cells for portable applications. J Power Sources
2002;106:3134.
8. Chan SH, Nguyen N-T, Xia Z, Wu Z. Development of a
polymeric micro fuel cell containing laser-micromachined
flow channels. J Micromech Microeng 2005;15:231236.
434
9. Ferrigno R, et al. Membraneless Vanadium Redox Fuel

Cell uisng Laminar Flow. J Am Chem Soc 2002;124:
1293012931.
10. Shah K, Shin WC, Besser RS. Novel microfabrication
approaches for directly patterning PEM fuel cell membranes.
J Power Sources 2003;123:172181.
11. Yu J, Cheng P, Maa Z, Yi B. Fabrication of a miniature
twin-fuel-cell on silicon wafer. Electrochim Acta 2003;48:
15371541.
12. Wainright JS, Saniell RF, Liu CC, Lit M. Microfabricated fuel
cells. Electrochem Acta 2003;48:28692877.
13. Barton SC, Gallaway J, Atanassov P. Enzymatic Biofuel Cells
for Implantable and Microscale Devices. Chem Rev 2004;104:
48674886.
14. Heller A. Miniature biofuel cells. Phys Chem Chem Phys
2004;6:209216.
15. Roundy SJ. Energy Scavenging for Wireless Sensor Nodes
with a Focus on Vibration to Electricity Conversion. Mechanical Engineering. Berkeley: The University of California;
2003 p 287.
16. Glynne-Jones P, Beeby SP, White NM. Towards a piezoelectric vibration-powred microgenerator. IRR Proc-Sci Meas
Technol 2001;148:6872.
17. Lu F, Lee HP, Lim SP. Modeling and analysis of micro
piezoelectric power generators for micro-electromechanicalsystems applications. Smart Mat and Structure 2004;13: 57
63.
18. Shenck N, Paradiso JA. Energy Scavenging with Shoemounted piezoelectronics. IEEE Micro 2001;21:3042.
19. Boland JS, Messenger JDM, Lo HW, Tai YC. Arrayed liquid
rotor electret power generation. Presented at IEEE MEMS
2005, Miami(FL) 2005.
20. Roundy S, Steingart D, et al. Power Sources for Wireless Networks. Presented at Proc. 1st European Workshop on Wireless
Sensor Networks (EWSN04), Berlin(Germany). 2004.
21. Roundy S, Wright PK, Rabaey J. A study of low level vibrations as a power source for wireless sensor nodes. Comp
Commun 2003;26:11311144.
22. Mitcheson PD, et al. MEMS electrostatic micropower generator for low frequency operation. Sensors and Actuator A
2004;115:523529.
23. Miao P, Holmes AS, Yeatman EM, Green TC. MicroMachined Variable Capacitors for Power Generaton. Presented at Electrostatics03, Edinburgh(UK). 2003.
24. Arakawa Y, Suzuki Y, Kasagi N. Micro seismic power generator using electret polymer film. Presented at The Fourth
International workshop on Micro and Nanotechnology for
power generation and energy conversion applications Power
MEMS 2004, Kyoto(Japan). 2004.
25. Li WJ, et al. A micromachined vibration-induced power generator for low power sensors of robotic systems. Presented at
World Automation Congress: 8th International Symposium
on Robotics with Applications, Hawaii 2000.
26. Mitcheson PD, Green TC, Yeatman EM, Holmes AS. Architectures for Vibration-Driven Micropower Generators. J Microelectomech Systems 2004;13:429440.
27. Ching NNH, et al. A laser-micromachined multi-modal resonating power transducer for wireless sensing systems. Sensors Actuator A 2002;97:685690.
28. Qu W, Plotner M, Fischer W-J. Microfabrication of thermoelectric generators on flexible foil substrates as a power
source for autonomous microsystems. J Micromechan Microeng 2001; 11:146152.
29. Wang W, Jia F, Huang Q, Zhang J. A new type of low power
thermoelectric micro-generator fabricated by nanowire array
thermoelectric material. Proc 22nd Int Conf Thermoelectrics
2003;682684.
30. Wenming Y, et al. Effect of wall thickness of micro-combustor

on the performance of micro-thermophotovoltaic power generators. Sensors Actuator A; in press, 2005.
31. Yang WM, et al. A prototype microthermophotovoltaic power
generator. Appl Phys Lett 2004;84:38643866.
32. Pescovitz D. The power of small tech. Smalltimes 2002; 2.
33. Lee SJ, et al. Design and fabrication of a micro fuel cell
array with flip-flop interconnection. J Power Sources
2002;112: 410418.
34. Chen T, et al. A Miniature Biofuel Cell. J Am Chem Soc
2001;123:86308631.
35. Moore CM, Minteer SD, Martin RS. Microchip-based ethanol/
oxygen biofuel cell. Lab Chip 2005;5:218225.
36. Bates JB, et al. Thin-film lithium and lithium-ion batteries.
Solid State Ionics 2000;135:3345.
37. Pique A, et al. Rapid prototyping of micropower sources by laser
direct write. Appl Phys A Mat Sci Proc 2004;79:783786.
38. Lee KB, Lin L. Electrolyte based on-demand disposable microbattery. Presented at IEEE MEMS 2002, Las Vega. 2002.
39. Applied Digital solutions, www.adsl.com.
40. Biophan technologies. Available at www.biophan.com/
biothermal.php.
41. MacDonald SG, Biothermal power source for implantable
devices. US Patent 6,640,137, 2003.
42. Whalen S, et al. Design, Fabrication and testing of the P3
micro heat engine. Sensors Actuator A 2003;104:290298.
43. Williams CB, Yates RB. Analysis of a micro-electric generator
for microsystems. Sensors Actuator A 1996;52:811.
44. Glynne-Jones P, et al. An electromagnetic, vibration-powered
generator for intelligent sensor systems. Sensors Actuators A
2004;110:344349.
45. Kang J-Y, Kim H-J, Kim J-S, Kim T-S. Optimal design of
piezoelectric cantilever for a micro power generator with
microbubble. Presented at Microtechnologies in Medicine
& Biology 2nd Annual International IEEE-EMB Special
Topic Conference, Madison (WI). 2002.
46. Kang JY, Kim JS, Kim HY, Kim TS. Micro Power Generator
with Piezoelectric Cantilever Driven By Micro Bubble. Sensors Actuator A submitted, 2005.
47. Meninger S, et al. Vibration-to-Electric Energy Conversion.
IEEE Trans VLSI Systems 2001;9:6476.
48. Peirs J, Reynaerts D, Verplaetsen F. A microturbine for electric
power generation. Sensors Actuator A 2004;113: 8693.
49. Miki N, Teo CJ, Ho LC, Zhang X. Enhancement of rotordynamic performance of high-speed micro-rotors for power
MEMS applications by precision deep reactive ion etching.
Sensors Actuator A 2003;104:263267.
50. Lee DH, et al. Fabrication and test of a MEMS combustor and
reciprocating device. J Micromech 2002;12:2634.
51. Backhaus S, Swift GW. A thermoacustic Stirling heat engine.
Nature (London) 1999;399:335338.
52. Frechette LG, et al. Demonstration of a microfabricated highspeed turbine supported on gas bearings. Presented at
Solid-state sensors and actuator workshop, Hilton head
island, (SC). 2000.
53. Jacobson SA, et al. Progress toward a icrofabricated gas
turbine generator for soldier portable power applications.
Presented at 24th Army Science Conference, Orlando (FL).
2004.
54. Das S, et al. Multi-Watt electric power from a microfabricated
permanent magnet generator. Presented at IEEE MEMS
2005, Miami(FL). 2005.
55. Harb J, LaFollete R, Selfridge R, Howell L. Microbatteries for
self-sustained hybrid micropower supplies. J Power Sources
2002;104:4651.
See also BIOTELEMETRY;
COMMUNICATION DEVICES; MICROFLUIDICS.
MICROSCOPY AND SPECTROSCOPY,

NEAR-FIELD
MIODRAG MICIC
MP Biomedicals LLC
Irvine, California
INTRODUCTION
The explosion of knowledge in life sciences is enabled by the
ability to visualize beyond the capability of the human eye
and by the capability to identify chemical compositions and
the structure of matter. This was enabled by Levenhooks
discovery of the new device for looking at the world of the
small: the microscope. He found that by combining two
lenses, it was possible to see much smaller objects than by
the naked eye alone. This led to his subsequent discovery of
the cell, which has spurred an explosion of knowledge in
life science and medicine that continues at a dramatic rate
of growth even today. Even 400 years after the Levenhook
discovery, one of the first tools of choice for the visualization of small objects is the optical microscopy. What is
known to a lesser extent is that the microscopic histochemical studies (i.e., staining of the tissues) with tissues and
organelle-specific dyes in the late 1800s, initiated the
modern pharmaceutical industry, when chemists and histologists alike envisioned an opportunity to selectively
inhibit or kill pathogens by organic molecules in the same
way organic dyes selectively label certain types of tissues,
cells, and organelles. While the optical microscopy methods
were able to uncover morphology and the structure and
nature of the cells, they were faced with the ultimate
physical limit of magnification, which is dictated by the
spatial resolution limited by the diffraction limit. This limit
was approximately the size of half of the wavelength of
light used to perform the imaging.
The breakthrough in imaging small structures and
further understanding the machinery of life and cell biology comes with the application of the deBroglies postulate
of particle-wave equality in order to use the electron beam
with a shorter associated wavelength as an investigative
imaging probe. Ruskas development of the first transmission electron microscope in late 1930s and the development
of scanning electronic microscopies in the 1950s, opened
the door to detailed investigations of the organization of
subcellular assemblies, viruses, and even imaging of the
individual biomolecules, at a resolution far beyond the
diffraction limit of visible light. The rapid advances in
the tools and techniques of ultramicroscopy, especially of
scanning probe microscopies, which for the first time
enabled routine molecular imaging, greatly contribute to
enabling completely new multidisciplines, like nanoscience
and nanotechnology, as well as an opening a door for an
entirely new way of looking into the machinery of life.
While scanning probe microscopy in the 1990s allowed
imaging at the unprecedented resolution of the unaltered
samples; in general, it lacked the ability to uniquely
identify the chemical composition of samples or unveil
their physicochemical properties. For the investigation of
chemical structures and fingerprinting the material composition, the tool of choice is optical spectroscopy. However,
435
the problem with classical optical spectroscopic techniques

is that it provides average, bulk results with no specific
information linking certain morphological features with
spectra. This can ultimately identify chemical composition
and/or physicochemical properties. The ability of doing
molecular fingerprinting and, in a raster pattern, subsequent molecular specific imaging at the nanoscale with the
spatial resolution of modern ultramicroscopy techniques is
the holy grail for many aspects of todays life sciences
disciplines.
This goal is partially fulfilled with electron microscopy
combined with energy-dispersive X-ray analysis (EDX),
wherein semiquantitatively, it is possible to associate topographical structures with elemental composition. However, for most of the problems in life and materials
sciences, simple knowledge of elemental composition is
not sufficient, as it is necessary to identify molecular
structure. Plus, the electron microscopy is, in most cases,
a destructive method of analysis, since the sample needs to
be prepared to be vacuum compatible, and be either electrically conductive, in the case of scanning electron microscopy, or have contrasts with heavier metals, in the case of
transmission or scanning transmission electron microscopy. Furthermore, the physics of generating characteristic X rays (i.e., the minimum size of the excitation volume
from which the signal is emerging) is in the tens of micrometers, thereby limiting elemental compositional analysis
with spatial resolution only for the large structure in the
tens-of-microns-sized range. The ideal technique will be
one that will allow imaging of the unaltered sample, in a
way similar to the way atomic force microscopy (AFM)
allows, while at the same time providing a way for spectroscopic identification of the chemical structure.
There are several techniques currently in their infancy
that promise spectroscopic probing with electromagnetic
spectroscopic information carrier signals imposed over
topography. They are near-field scanning optical microscopy, microthermal analysis, scanning nuclear magnetic
resonance (NMR) microscopy, and scanning electron paramagnetic resonance (EPR) microscopy.
However, for solving any of the practical problems, it
will be of great benefit that the ultrastructures information probes are photons of visible, and near-infrared (IR)
and ultraviolet (UV) light, as a great deal of both morphological (based on the photons position and intensity/count)
and compositional (based on adsorption, fluorescence,
Raman shift, etc.) information can be simultaneously
obtained as optical microscopy relies on light as an information carrier. This is due to the fact that the same
photons, which are in standard imaging configurations
used to generate images, carry much more information
on composition and the various physical and chemical
properties of the observed spot on the sample that can
be extracted through different spectroscopic methods.
The technique that has evolved over the last decade and
promises to fulfill the above-goals at the nanoscale level, is
near-field scanning optical microscopy (NSOM or SNOM),
which effectively breaks the physical limits imposed by
the optical-diffraction-limited resolution by using the
near-field evanescent waves and scanning mechanisms
similar to those in the scanning probe microscopies.
436
THEORETICAL PRINCIPLES OF NSOM MICROSCOPY

Abbes equation (Eq. 1) describes the resolution of the
classical, far-field optics, (i.e., the minimum separations
between two points that can be distinguished) (1). From this
equation it is easy to conclude that the maximum attainable
resolution in the far field is 12 of the applied wavelength,
which means that the best optical microscope cannot be
used to visualize details smaller than 200400 nm.
pmin
l
2n sin a0
In Abbes equation, n is the diffraction of the imaging

index and a0 is the aperture angle in the medium. While
Abbes equation describes the limiting resolution in the world
of conventional optics, the Fourier optics approach can provide us with the same conclusion. Following Abbes principle,
Rayleigh (2) derived that the objects in a lens system in the
far optical field are resolved only when the maximum of one
pattern coincides with the first minimum of the neighboring
features. What resulted was the discovery that the resolution
criteria that describe the maximum resolution of the optical
system based on the size of the numeric aperture was
d 0:61 l=NA
wherein l is the applied wavelength and NA is the numerical

aperture of the lens, again bringing the maximum theoretical
resolution to 200 nm.
A similar observation can be derived from the Fourier
formalism in optics. In Fourier optics, the information content embedded in the spatial frequency f, in the case when f is
higher then 1/l, decays rapidly toward zero from the object
and thereby, no data on the subwavelength features can be
efficiently collected with standard far-field optics. However, it
is well known that it is possible to receive an electromagnetic
signal with antenna that is smaller in size than the wavelength. The solution of the Maxwellian equations that govern
the behavior of electromagnetic radiation differs in the distance smaller than the wavelength than in the distance,
much larger than the considered wavelengths. When the
waves propagate within the distance much smaller than
its wavelength, such situation is called the near field. The
pragmatic definition of near-field optics will be a division of
optics that deal with the elements of the subwavelength
features scales, which are intended for passing the light
through, from or near, to another element with subwavelength features positioned within the subwavelength distances. The spatial resolution in near-field optics depends
on the features size and is limited to about one-half of the
aperture size. Furthermore, the near-field system must be
considered as a complete system consisting of two features
(probe and sample in the case of microscopy) and the resolution
of the system will be dependent on both sample and probe.
Thus, it is impossible to speak of the unique or standard nearfield resolution, as is done with a far-field instrument.
NEAR-FIELD IMAGING EQUIPMENT
The near-field scanning optical microscopy or scanning
near-field optical microscopy (NSOM or SNOM) is a tech-
nique that enables users to work with standard optical

tools that are integrated with scanning probe microscope
(SPM) technology to obtain the optical image at a resolution in the range of tens of nanometers. This is quite
comparable with the resolution of scanning electron or
SPM. The integration of scanning probe and optical methods allows for the collection of optical information at resolutions well below the optical diffraction limit, which
overlap real topography information obtained by scanning
probe feedback. For spectroscopy applications, NSOM
offers the potential for characterizing the spectroscopic
signature of material on a submicron-to-nanometer scale,
thereby affording new insights into nanoscopic structure
and composition.
The principles of NSOM microscopy were theoretically
founded by Synge in 1929 (2), and in his subsequent paper
he described an imaginary device that closely resembles
todays NSOM setup (3), including the use of piezoactuators. Due to technical difficulties at the time to implement
such a device, the idea was forgotten until theoretician
OKeefe rediscovered the idea in 1956 (4). The first practical demonstration of the imaging of a structure smaller
than the one-sixtieth of the applied wavelength was done in
1971 using the microwave in near-field scanning over the
grid (5). The basic idea behind this method was to create
evanescence, a standing wave, by light diffraction through
an aperture that was much smaller than the wavelength,
and then to use this evanescent light source to scan a
sample in a raster-scan pattern in close proximity, and
collect transmitted or reflected signal in the far field. The
first demonstration of near-field optical imaging was implemented independently by Pohl (6) in 1982 and Lewis
groups (7) in 1983, and described as an optical stethoscopy,
in what is now considered the beginning of NSOM microscopy. The method grew rapidly during the 1990s and the
trend is continuing to this day, as described in recent
reviews (812). Furthermore, several companies are offering commercial instruments (1316) that enable ordinary
users, who are not inclined toward the instrumentation
development, to apply NSOM in solving their research
problems.
There are two fundamentally different ways to achieve
near-field optical imaging. They are apertured-base and
apertureless NSOMs with their principles of operation
depicted in Fig. 1a and b (17). In the case of the apertured
NSOM (Fig. 1a), the light passes through an aperture that
is much smaller than the wavelength of applied light, and
ultimately the resolution is defined by the size of the
aperture. In the case of the apertureless NSOM (Fig. 1b),
a sharp metallic tip is irradiated by a laser perpendicular
(or as close as possible) to the tip along the axis. The
irradiation excites the plasmons on the metallic surface of
the tip and the field is concentrated by combining antenna
and plasmonic effects at the top of the tip. The resolution
of apertureless NSOM is thus defined by the size of the
near-field excitation formed at the apex of the metallic
tip, and is determined by tip sharpness, tip materials, and
real and imaginary parts of the refraction index of the
used metal.
The practical advantage of apertured NSOM is in its easy
implementation. While the advantage of the apertureless
437
Figure 1. Two major principles of achieving the nearfield scanning optical microscopy: (a) the aperturebased NSOM; (b) scatter-field, apertureless NSOM.
NSOM is theoretically a higher achievable resolution and

possibly higher field strength, the technical difficulties in
implementing the apertureless setup have not permitted
those advantages to be realized. Thus, as of this writing, all
of the existing commercial instruments are based on the
apertured NSOM approach.
Regardless of the type of NSOM, the experimental setup
always consists of a piezo XY scanner, whose role is to
execute the raster-scan pattern scanning of the sample by
the near-field probe; a Z piezo, whose role is to modulate
the image by keeping the sample-tip distance; an optional,
but for most cases necessary, noncontact modulation element, such as the tuning-fork for shear force feedback, or a
piezo- or electromagnetic oscillator for AFM-like noncontact feedback; a near-field probe, which can be aperture or
sharp tip; a laser light source; a far-field optical signal
collection system, sample and sample holder; a system for
coarse probe approach and sampleprobe alignment; a
scanner controller and computer for the image acquisition
and reconstruction; and a vibration isolation system. In
this manner, the NSOM most closely resembles the
mechanism of the AFM, and almost all of the existing
commercial setups, as well as many of the in-house, labmade NSOMs, share these common components with the
AFM and more general SPM platforms.
Piezo Scanner
The piezo scanner principle of work is based on the piezo
effect, which is reversible internal stress induction within
the crystal when exposed to the electric field (18). This
stress induces crystal expansion. The piezo scanner in the
NSOM is a more critical part than in the standard AFM.
Ideally, it should be the perfect closed-loop scanner, due to
stringent requirements of keeping the probe in a particular place during the raster scan, in order to achieve
sufficient optical signal/noise ratio. Figure 2 depicts typical scanner configurations. The scanners are usually
implemented in the form of the stacked piezo crystals
(Fig. 2a), tube scanners (Fig. 2b), and bimorph (Fig. 2c)
(19). Furthermore, scanners are often grouped into the
so-called tripod configuration. The same material used for
the SPM scanner, lead-zirconate-titanate ceramic (commonly referred as a PZT ceramic), is commonly used for
the NSOM piezo scanner. The typical piezo-electric constant for PZT materials is about 1.7 V nm1. However,
in order to practically achieve the linearity over the whole

range of scan, it is necessary to calibrate each individual
scanner periodically to compensate for crystal nonlinearity,
creeping, and drift. Those effects are further minimized by
using active, real-time feedback, which can be implemented
either through some form of the strain gauge, capacitance,
or by optical means. The active feedback adjusts the
voltage applied to the scanner to keep linearity and to
secure the probe above the scanning position within the
raster scan.
Optical Signal Acquisition System
The optical signal collection system is made up of optical
and optoelectronic parts. The optical portion usually consists of the far-field microscope objective with a high NA
lens. The tip-sample working distance, as well as the
sample thickness and sample holder accessibility, define
the maximum NA of the objective that can be used. Oil
immersion objectives are used to enhance the NA by many
times. Besides the objective, the collection system may
contain filters, notch-filters polarizers, and beam splitters,
depending on the particular configuration and imaging
mode. The optoelectronic part of the collection system
converts optical information to an electrical signal for
further processing. It is usually either a highly sensitive
photomultiplier tube (PMT), or, for ultimate sensitivity
and single-photon counting, an avalanche photodiode
detector (APD) array. For the PMT tube, the output signal
can be either voltage or counts, and for the APD, it is only
TTL (transistortransistor logic) counts. The high sensitivity PMT tubes can satisfy most of the imaging requirements, however, for extremely weak signals, such as in
single-molecular imaging or single-molecular spectroscopy, an APD detector is more desirable. Due care does
need to be paid when using the APD detector, as overexposing the detector can damage it in an extremely short
period of time. For the purpose of correlated experiments,
many detectors are attached to the system. In the case of
spectroscopy applications, the most commonly used dispersive detector is a highly sensitive CCD imaging camera,
either solid-state or liquid-gas cooled. However, many setups use the other, more economical, wavelength or energydispersive detection systems, which are based on filters,
prisms, or gratings in conjunction with either a PMT or
APD detector.
438
Figure 2. Schematics representation of the piezo actuators: (a)

stacked piezo; (b) tube scanner; (c)
bimorph scanner. [Courtesy PI
(Physik Instru-mente) LP, www.pi.ws.
Light Sources for Near-Field Imaging and Spectroscopy

Illumination sources for near-field microscopy are always
lasers. Selection criteria for the laser depends on the
desired wavelength(s). The most economical are the
solid-state lasers, followed by ion lasers, such as the Ar
laser, which can selectively produce multiple wavelengths
of light from 450 to 514 nm (17). Besides the two most
common types, many setups use liquid lasers as well as
optical parametric oscillators to produce specific wavelengths that are not available with a standard laser. The
transduction path can consist of mirrors or single-mode
optical fibers. The mirror-based path has better throughput, however, it is more complex to adjust and requires
periodic readjustment. The optical-fiber-based transduction path has some higher attenuation then the mirrorbased path, but it is very convenient for use, especially if it
is implemented with the standard FC or similar connectors.
Microscope Head
The NSOM head usually consists of a probe holder, one or
more piezo scanners, a system for coarse probe approach, and
a case. The head is positioned at the top of the sample holder.
The probe holder is directly attached to the Z direction piezo.
A video system, which shows the image of the approaching
tip and substrate, and a software- controlled stepper motor
with micrometric screws provide coarse approach of the probe
to the sample surface. When the probe tip is brought to close
proximity to the sample, the fine approach mechanism is
engaged. The fine approach mechanism is essentially a
stepped mechanism wherein the probe is brought in a small
piezo steps in the vertical direction to close the gap between
the tip of the probe and sample substrate. The contact is
achieved when the signal indicates the deflection of the probe
in AFM-like setups, or, in shear-force mode, when its interaction with the sample reaches the user-prescribed set
point voltage or current level. The determination of the
appropriate set point varies for different samples and systems, and is more a result of art or tacit knowledge than an
exact science. If the Z piezo is completely extended and
contact has not been achieved, the piezo constricts to its
neutral position and the stepper motor is activated to bring
the probe to the approximate max extension distance of the
piezo, and the process is repeated. Besides the Z piezo,
sometimes the XY scanners can be positioned in the head.
Sample Holder/Stage
The sample holding stage can contain the XY piezo scanner, if it is not in the head. Its moving frame consists of
micrometric screw positioners that push the sample holder
in the XY direction under the probe, thus allowing sample
pan operation. These screws can be manually or steppermotor operated. The sample stage can be stand-alone, or it
can be positioned at the top of an inverted, epi-fluorescence
microscope. There are many advantages to having the
NSOM sitting at the top of the standard inverted microscope. This configuration is able to combine far- and nearfield microscopy, exploit the operation familiarity of the
inverted fluorescence microscope, and deliver superior
images to those acquired via a dedicated, stand-alone
NSOM stage. However, a drawback of such a configuration
is a larger mechanical circuit with a higher level of vibrational noise than in a dedicated system.
Controller
Controllers for NSOM are usually derived from the AFM/
SPM controllers. In all of todays setups, they are digital.
The controllers role is to generate the high voltage signals

necessary to feed the piezo scanner and move the probe
in the raster scan pattern. It also controls the vertical,
Z position of the probe via the PID control-loop model
mechanism (proportional-integral-differential), and thus
topographically modulates the signal; it maintains
the noncontact feedback; it controls the coarse probe
approach, and in some instances coarse sample positioning;
and it acquires the signals coming from the probe (both
optical and topographical) and forwards them to the computer for further processing. The controller usually consists of a series of analog-to-digital and digital-to-analog
converters, precision operational amplifiers, and high
voltage amplifiers. Due to the complexity of the tasks, often
the controller is designed using high end digital signal
processors and other high end embedded systems.
The role of the control software is to control the controller, acquire the image, and store it in some of editable
and exportable format. Furthermore, the control software
almost always possesses image processing capabilities,
such as different image filters, Fourier transform, 3D
representations and rendering, and so on. All of the commercially available NSOMs share the same software with
their AFM/SPM cousins. Many of the homemade systems, on the other hand, have software modified from
existing commercial SPM/AFM controller software, or software that is independently written, as in cases where the
users have designed the whole control electronics by themselves. Many times, the results are processed in third party
software. For example, for image processing a very popular
solution is to use shareware NIH Image software or its PC
cousin, Scion Image, and for advanced applications, to use
scripts written for IgroPro, LabView, MathLab, or for
spectroscopy experiments, WinSpec. Use of higher level
software like Igor Pro dramatically reduces development
time of applications, as compared to the time required to
write the script in C or C code.
Apertured NSOM
The principle of the apertured NSOM is to use the aperture
as a scanning probe. This is the first (5,6) and up-to-today
most commonly implemented NSOM setup. In basic principle, the instrument consists of the XYZ piezo scanner(s)
that moves the apertured probe over the raster scan pattern at the controlled aperture-sample height, and a noncontact feedback mechanism, the best-suited being the
shear-force based one (20). An aperture in the tens of
nanometer size can be formed by the tapering, heating,
and pooling process borrowed from biophysics labs, where
it is utilized for creating micropipettes (21) or for etching
optical fibers (22). Additionally, as an aperture, it is possible to use the hollow cantilever (23).
Schematic representation of the instrument implementation, for both imaging and spectroscopyhyperspectral
imaging, is presented in Fig. 3a, configured for the most
common, transmission mode (looking through the sample)
operation. The laser light is coming from the optical fiber.
Optical fiber is mounted on the tuning fork assembly,
which is held onto the Z-piezo scanner and is constricted
at the end to a tens of nanometer size range (see insert) and
439
Figure 3. Example of the aperture-based, straight-fiber NSOM:

(a) schematics representation of the scanning probe and collection;
(b) schematics representation of the tapered fiber NSOM probe,
attached to the tuning fork; (c) photography of the probe glued to
the tuning fork; (d) SEM image of the end of the tapered,
aluminum-coated fiber optic based NSOM probe. (Courtesy of
Veeco Instruments Inc., Santa Barbara, CA.)
the evanescent field is formed at its aperture, as represented in the figure insert. The light is passing through the
sample, interacting with sample matter, and is collected
under the sample in the far-field with the high numerical
aperture microscopy objective. Such a signal is further
subjected to collection and processing. For the hyperspectral
imaging or for the spectroscopy or spectral imaging purposes,
the signal is first passed through the holographic notch
filter, which eliminates excitation light and through the
beam splitter is directed to the wavelength-dispersive detector,
such as a CCD spectrometer, and to the summary, imaging
detector, such as a PMT, or avalanche photodiode detector
(APD). In the simplified setup, if the apparatus is used just
for the optical imaging, the light is passed directly from the
objective into the imaging detector, (i.e. APD or PMT).
Figure 3b schematically represents a typical apertured
probe, mounted on the tuning fork. Micrographies at
440
Fig. 3c and d represent the frontal and lateral view of the

metallic-coated, laser-pulled, tapered, fiber-based tip. An
evanescent, standing wave is formed at the end of the
aperture, and the size of the aperture approximately
defines the optical resolution. Light is either brought
through the aperture, as in transmission and reflection
imaging mode, or collected through the aperture, as in the
collection mode. The tip is scanned across the sample in a
raster-scan pattern, and for imaging purposes; the signal is
collected in the far field, either by sensitive photomultiplier
tube, or by sensitive avalanche photo-diode counter.
The three distinctive different modes of operations of
the aperture-based NSOM are illustrated in Fig. 4, which
also graphically depicts the different kinds of information
that can be extracted from the optical signal emanating
from the sample. The origin of the optical contrast, as
depicted in Fig. 4, can be due to topographic differences
(different path length change the adsorption), material
birefringence, reflectivity, sample extinction coefficients
for the particular excitation wavelength, index of refraction, fluorescence emission properties, nonlinear spectroscopical properties of materials, and mechanical and
magnetic stress in the sample. However, at the moment
of this writing, for life sciences and biomedical applications, only transmitivitty, reflectivity and fluorescence
properties are of significance. The mode that is the most
useful for biological applications is the transmission mode
or the looking through mode, and is most similar to
classical biological microscopy In this case, as described
above, light is brought through the fiber-based tip, the near
field interacts with the sample, and the signal is collected
in the far field as it passes through the sample. In this
mode, it is possible to do transmission imaging, as well as
fluorescence or other wavelength-resolved imaging, by
application of adequate filters or wavelength-selective elements. In a reflection mode, which can be described as
looking on the surface mode, the near field interrogates
Figure 4. Typical apertured NSOM configuration: (a) illustration

of information that is carried and can be extracted from the optical
signal; (b) transmission and reflection mode NSOM; (c) collection
mode NSOM; (d) total internal reflection or dark field mode.
(Figure 3bd adapted from Ref. 17.)
the surface of the sample, and the scattered signal is

collected in the far field. This mode allows imaging and
spectroscopy of the nontransparent samples; however, the
imaging efficiency is much lower than with transmission
mode. In a collection mode, the light is passed either
through the sample, or illuminated on the sample, and
the signal is collected through the fiber in the near field.
This mode is very burdensome to use, has low signal
collection efficiency, and is used mainly in photonics
research.
Besides these three modes, there are more exotic modes
of operation, such as combined collection and illumination,
where both sample illumination and resulting signal are
passed and collected through the same probe; dark field
imaging, where the probe tip is in close proximity to a
sample that is illuminated from underneath with total
internal reflection from the substrate, and wherein the
probe acts as a second, tunneling prism. Besides pure
optical operation modes, there are also a combined optomagnetic NSOM, which explores Kerrs effect (24); nanomass spectroscopy (25), where near field is used for ablation; and optoelectrochemical NSOM (26). The later two
modes have a lot of potential applications in physiology
with their ability to simultaneously image and record
potential at subwavelength resolution.
To secure the probe in a near field, the aperture must be
kept in close proximity to the sample with a distance much
smaller than the applied wavelength. The fiber is kept at
the nanometer-range distance from the sample by means of
noncontact feedback. There are several ways of achieving
feedback, mainly shear force, AFM-like normal force contact, and noncontact force feedback. The shear-force feedback provides gentler, lateral touching of the sample,
thereby reducing the possibility of aperture contamination
or tipaperture mechanical failure.
In shear-force feedback (20), the fiber tip is oscillated
laterally to the sample surface. The NSOM tip is rigidly
premounted on a quartz tuning fork (Fig. 3b and c), which
is a few millimeters in size. The tuning fork is mechanically
vibrated at resonance frequency, usually in tens to hundreds of kilohertz, resulting in a few nanometers of lateral
motion at the distal end of the NSOM tip. When the tip is in
a close lateral proximity to the sample, the resonance
frequency of the tip-tuning fork system is disturbed due
to electrostatic, van der Waals, hydrogen bonding, and
other kinds of attractive and/or repulsive interactions
between the tip and the sample. This disturbance is read
as an electrical signal that is processed, and the tip is
moved accordingly in the vertical direction to achieve its
preset resonance frequency, thus keeping the same distance from the sample.
Optical resolution, which is typically achieved by fiberbased apertured NSOM, is in the range of 50 nm, with
maximum resolution being in the range of 20 nm. The
improvement in tip fabrication procedures and in the control of the tip-sample separation distance will ultimately
lead to better resolution. For apertured NSOM, fiberoptic
or pipette-based tips are fabricated by constricting the core
of the optical fiber to a 5020 nm diameter. This is achieved
by a heatingpulling method (21), wherein the fiber is
transversally irradiated by CO2 laser and simultaneously
stretched on the pipette puller until the fiber is broken, or

by chemical etching (22) at the phase boundaries using the
HF solution with oil on the top. To enhance the efficiency of
the light transmission and to avoid the light leakage
through the fiber shell, the probes are usually coated with
either aluminum or silver. The coating is done by vacuum
evaporation, and its role is to prevent light leaking out of
the probe. The metallic coating is especially beneficial in
the near-field surface-enhanced Raman spectroscopy.
Another way of performing apertured NSOM is by
bending the fiber or pipette. In this case, the force feedback
can be achieved either by shear force, or by AFM-like
normal force in both contact or noncontact mode. The
disadvantage of this approach is that such bended fibers
are more vulnerable to mechanical failure, and if used for
spectroscopy purposes, there may be problems with Raman
scattering lines coming from the fiber shell materials.
The other way of achieving apertured NSOM imaging is
by using the hollow AFM cantilevers (23). In this kind of
setup, the light from the excitation laser is focused on the
top aperture on the center of the hollow AFM tip, and the
near field is formed at its bottom. The feedback mechanism
used therein is the same as in noncontact AFM. Presently,
the resolution (in the range of 100 nm) of such hollowcantilevered-based apertured NSOMs is inferior to that
of pulled-fiber-based NSOMs. Another disadvantage of the
AFM-like force-feedback setup in NSOM applications is in
that the AFM uses a laser beam to follow the bending of the
cantilever. In NSOM, when many applications are counting the individual photons, the optical noise introduced by
the AFM-like laser-based feedback may be several folds
stronger than the signal. Considerable improvement is to
be expected with piezo-actuated hollow cantilevers, which
will avoid using laser feedback.
Apertureless NSOM
In the apertureless NSOM (Figs. 1b and 5), a sharp metallic
tip is irradiated by the laser light orthogonally to the long tip
axis, and the near-field excitation is scattered from the
Nanoscope
Controller
Computer
Scanner
Controller
Computer
441
tip (27,28). The light scattering from the feature is much

smaller than the applied wavelength, which also generates
the strong evanescent field. The best strength of the scattering field is achieved if the excitation laser frequency
corresponds to the surface-plasmon resonance of the metal
from which the tip is made. Incoming beam scattering
produces the evanescent field at the tip; however, the
physics of the process is a combination of the near-field
antenna effects and surface plasmon resonance. The laser
induces the plasmons in the tip, which oscillates in parallel
to the tip axis and amplifies the evanescent standing wave
at the tip apex. The standing wave interacts with the
sample and the signal is collected either in transmission
or reflection mode in the far field. The tip is scanned across
the sample in the same raster-pattern manner as with
apertured NSOM. Feedback is provided in either the
noncontact AFM manner, preferably with a tuning fork
or some other nonlaser based Z-deflection feedback, and
the signal collection is modulated by oscillating the probe
in a vertical direction in order to avoid static and scattering
artifacts. Furthermore, in modulated apertureless NSOM,
the signal from the photodetector is also modulated with the
same modulation signal source as a tip, in exactly the
same frequency and phase (with possibilities of higher
harmonics modulation. This is done in order to avoid
inbound laser light nonnear-field induced scattering;
static-scattering artifacts from sample features and to achieve
optical signal acquisition always in a same sample-tip
separation distance position.
In order to distinguish between the near-field scattering
and inbound laser light, most of the apertureless NSOMs
are used mainly for fluorescence, Raman, or for different
nonlinear optical phenomena applications. Furthermore,
because of the rapid decoy of the scattered field, modulation
of the scanning probe, and control of the sample-tip separation distance in the apertureless configuration is much
more critical than in the aperture-based NSOM.
Figure 5 is a schematic representation of a typical,
homemade, apertureless NSOM setup; in this particular
case used for fluorescence and fluorescence-lifetime (FLIM)
AFM head
Laser
Coated AFM tip

X-Y Piezo
Stage
Microscope
Objective
Beam
Splitter
Filter
APD
Filters
SPC730
Computer
Figure 5. Schematic representation of the example

of the apertureless NSOM. (Adapted from Ref. 29.)
442
near field imaging. It consists of a commercially available

AFM head, mounted on the top of the inverted epifluorescence microscope, positioned at the optical table.
The lateral irradiation of the tip, which is necessary in order
to achieve the high intensity evanescent field generation, is
produced by offsetting the position of the AFM tip in
relationship to the high numerical apex microscope objective. In this particular case, the fluorescence imaging
measurements were conducted in an inverted fluorescenceimaging microscope (Nikon Diaphot 300), the excitation
light from a mode-locked YAG laser (Coherent Antares)
at 532 nm wavelength, 10 ps pulse width, and 76 MHz
repetition rate was focused on a diffraction-limited spot
through an objective (Nikon 60X NA 1.4), and the emission
from the sample was collected by the same objective, in
the epi-fluorescence manner. The emission band-pass
filters were HQ565/25 and D570/20 (Chroma Technology)
to ensure that the excitation light and the feedback
laser of AFM (650 nm) were both blocked. The emission
was detected by an avalanche photodiode (APD) (Perkin
Elmer, SPCM-AQR-15). The background photon counts
with AFM feedback on were 150 Hz. The sample cover
slip was mounted on a closed-loop two-dimensional (2D)
piezoelectric scanner (Polytec PI, P-731). The AFM (Veeco
Instruments Inc, D3100) head and inverted microscope
were coupled at an overunder position. The AFM tappingmode tips used in this work are commercially available Si
tips (Digital Instrument, OTESP7) coated with Au and Ag,
by sputter coating. Image density of 128 128 pixels and
scan rate of 1 Hz. As the quenching effect is highly distance
dependent, the tip oscillation amplitude was reduced by
reducing the driving voltage to the tip as much as possible
without sacrificing image quality. Based on the force calibration curve, the tip oscillation amplitude during the
imaging was estimated at 30 nm.
The sample-scanning confocal fluorescence image was
recorded by a home-built computer control interface that
counted the APD signal and raster-scanned the piezoelectric scanner. The fluorescence decay traces were
recorded by a time-correlated single photon counting
(TCSPC) module (Becker & Hickl SPC730, Germany).
The start signal was from the APD and the stop signal
was from synchronization of the YAG laser at one-half of
the laser repetition rate. For the lifetime imaging mode,
the TCSPC module reads the line-synchronization signal
of the Digital Instrument Nanoscope IIIa controller to
achieve a synchronized recording of the AFM signals
and fluorescence signals.
Figure 6 depicts the FEM simulation (30) of near-field
enhancement around a metallic tip positioned in close
proximity to the sample and irradiated with a laser beam.
Figure 6 shows the rapid dependence of the near-field
excitation on the sample-tip separation, and emphasizes
necessity of accurate, sub nanometer sample-tip separation control mechanism for any widespread, commercial
applications. This is even more important for the Raman
spectroscopy or hyperspectral imaging (or in this sense for
any other, nonlinear optical applications), as the strength
of the Raman emission is proportional to the fourth power
of the strength of the electric field of applied light the small
changes in the strength of the local near-field enhancement
Figure 6. Finite-element methods (FEM) simulation of the

electromagnetic field scattering and near field enhancement at
the apex of the tip of the apertureless NSOM, and its behavior with
change in tip-sample separation distance. (Adapted from Ref. 30.)
can produce the dramatic fluctuation in the strength of the

optical signal.
Besides all of the difficulties in implementation, the
advantage of the apertureless mode is in theoretically
better resolution than the apertured mode, and in the
higher surface-enhanced Raman signal, due to plasmon
coupling, which makes this method, theoretically, an ideal
molecular Raman nanoprobe, a holy grail for life scientists (31,32). The hyperspectral subwavelength Raman
imaging is extremely important for further studies in
system biology, proteomics, and metabolomics, as it is
expected it will for the first time allow identification and
spatial positioning of biomolecules within a cell, without
the introduction of fluorescence labels. Aside from Raman
imaging, this approach is expected to be better for the
purpose of fluorescence lifetime imaging (FLIM) (29).
The apertureless approach also has significant advantages
because the surface plasmon enhancement is driven by the
metallic tip and a higher local intensity of the scattered
field.
However, to date, there is no commercial instrument
available based on the apertureless NSOM principle,
because of many technical problems, a significantly smaller
photon flux, a low signal/noise ratio, and extreme signal
dependance on the tip-sample separation distance. It is
expected that with improvements in the control mechanism
for keeping the sample-tip separation in the subnanometer

range, as well as improvements in laser positioning and
further enhancement of the photodetector efficiencies, the
apertureless NSOM will become more widespread, with the
first commercial instruments expected to be introduced to
the market in the near future.
The tip for the apertureless NSOM can be manufactured
in three different ways. The simplest way is by electrochemically etching the metallic wire in the same way as for
production of the STM tips. However, controlled and optimized shapes, which can provide more efficient field
enhancement, can be better achieved by using the free
ion bombardment (FIB) techniques for both tip growth
and etching.
NSOM Operations
The typical operation of the NSOM consists of positioning
the tip above the sample feature of interest, visually
using the reflection and transmission video system, respectively. After executing a manual approach procedure, the tip
is subsequently placed under PID control and is automatically maintained in the near-field region. The nonoptical,
shear-force feedback relies on measuring the voltage generated by a quartz tuning fork onto which the NSOM tip is
rigidly mounted, thus avoiding feedback laser. Having a
feedback without the feedback laser is of great advantage, as
avoidance of that voltage is a direct measure for the oscillation amplitude of the tip-tuning fork assembly, which varies
with tip-to-sample distance over a range of 25 nm.
Unlike conventional AFM cantilever designs, the spring
constant of the straight-fiber NSOM tip in the vertical
direction is extremely high, thus avoiding damaging
snap-to-contact. A feedback algorithm monitors the amplitude of the tuning fork by appropriately adjusting the tipsample distance. Using this method, the NSOM tip is
engaged and maintained within 5 nm of the surface in
the near-field region throughout the NSOM scanning or
spectroscopic measurements. Another advantage of using
shear-force feedback is in the absence of a feedback laser,
which is especially important in the low photon-count
applications (e.g., in spectroscopyhyperspectral imaging
and single-molecular studies).
Other methods of maintaining the tip in the near field
have not proven nearly as sensitive or reliable as tuningfork-based shear-force feedback. Some methods originally
developed for AFM applications may require actual surface
contact and, consequently, possible surface or tip transformation, ultimately resulting in either damage or having to
move the tip in and out of the near-field during data
acquisition. The other disadvantage of AFM-like feedback
is in the great technical difficulties to form a self-actuated,
piezo-based AFM hollow tip, thus forcing the use of laser
for feedback control. The AFM-like force feedback with
pulled fiber tip requires a bent fiber, which is much less
mechanically stable than a straight fiber. In addition, the
bent fiber has problems associated with circular light paths
and higher Raman scattering from glassfiber substrates,
interferences that carry especially negative consequences
for near-field spectroscopy, as they can significantly
increase the optical noise level.
443
The NSOM can be used both in air and in liquid. Most of

the work to date has been done in air. While it has been
demonstrated many times that the method can be successfully used in liquid operation, there are problems associated with the menisc force formed between the probe and
liquid surface. For work in air, shear force is the superior
method of feedback. However, for work in liquid, the AFMlike noncontact feedback has advantages. With further
improvements in the fiber-based probes coating in the near
future, it is expected that shear-force-based topographic
imaging and feedback will become equal with the AFM-like
noncontact-based topographic imaging in liquid.
Near-Field Spectroscopy
For spectroscopic studies, NSOM can be considered as a
controlled light collector, with tens-of-nanometer spatial
positioning resolution. Thus, many of the standard
spectroscopic techniques could be applied, depending
on the amount of signal available. For example, it was
successfully demonstrated that NSOM can be used for
fluorescence and photoluminescence spectroscopy; electroluminescence spectroscopy, time-resolved spectroscopy;
polarization studies, and in early stage infrared (IR) and
Raman spectroscopies. The latter of the two has the greatest promise in becoming the ultimate molecular nanoprobe.
Some of the applications of in situ hyperspectral, that is,
composition-specific nanoscale studies include chemical
identification of observed samples; studies of optical properties of materials at nano-scale levels; detection of phase
differences and impurities in materials; protein studies,
and many more. Ultimately, it opens the door for a plethora
of both fundamental and applied studies in the fields of
physics, material science, chemistry, life sciences, and
nanotechnology.
Examples of a near-field spectroscopy application are
shown in Fig. 7. While Fig. 7a represents the topography
image of the PIC dye crystal (33), Fig. 7b is the NSOM
fluorescence image of the same area. In order to explore the
origin of inhomogeneities, near-field fluorescence spectroscopy, with spectra presented at Fig. 7c, has been done at
different points, labeled 14, on Fig. 7b. Finally, fluorescence spectra resolved the inhomogenieties of emission
sources, pointing to the two different allotropes of crystals
having emissions peaking at 645 and 690 nm, respectively.
The near-field signal, which is by default very weak,
gets even weaker if we want to do the wavelength-resolved
spectroscopic analysis, or full hyperspectral cube imaging,
so longer exposition time is necessary to acquire usable
spectra. For a near-field spectroscopy system to be successful, it needs to have an extremely stable probe position
control, in all three axes, to keep the optimal sample-probe
distance, and to keep the probe above the point of interest,
for a prolonged time of signal collection.
Figure 8 represents the typical modern commercial
NSOM microscopyspectroscopy setup, in this particular
case, an Aurora-3 for spectroscopy made by Veeco Instruments Inc., Santa Barbara, CA (34). In general, such a
near-field microscopyspectroscopy package consists of the
NSOM microscope, an objective lens for signal collection,
optical filters for elimination of the excitation laser light,
444
Figure 7. Example of fluorescence-based hyperspectral near-field

imaging. (a) Shear-force topography image; (b) NSOM fluorescence
image; (c) spatially resolved fluorescence spectra, numbers 14
corresponds to the different position on the crystal, demonstrating
chemically specific imaging and material inhomogenities at the
nanoscale. (Courtesy of David Vanden Bout and Paul Barbara,
University of Minnesota, Minneapolis, MI.)
an optical pathway for signal transduction to the detector,

and a wavelength dispersive detector. As the signal is
generally very weak, the spectrometer needs to have a
very sensitive detection system, in the best case, singlephoton sensitivity. With todays solid-state detectors technology, the best detector to use is the CCD camera with a
larger stack of sensitive pixels coupled to the imaging
spectrometer. Depending on applications, either a Peltier
cooled camera will be satisfactory (for most bright fluorescence samples) or a liquid-nitrogen-cooled camera for ultimate sensitivity, such as in single-molecular experiments
and near-field Raman spectroscopy. Furthermore, interfacing and communication between spectrometer and NSOM
scanning probe control software needs to be established.
Today, many spectrometers have semiopen control software, thus allowing triggering of the spectral acquisition
with the TTL handshake signal, which can be produced
by the NSOM microscope controller. In this way, the
microscope controller initiates spectral acquisition and
the system can be used with many different commercial
spectrometers, allowing customization of the microprobe.
Filtering out excitation photons is of extreme importance to
increase the signal/noise ratio, and the best filters available today are notch, interferometric filters. Another consideration when designing the NSOM spectroscopy
package is that not one formula will fit all of the requirements. Furthermore, virtually any application will require
some special design consideration, so building the system
Figure 8. Typical, third generation commercially available NSOM setup, for imaging and
spectroscopy: (a) photo of the NSOM head and sample holder; (b) complete system with
attached spectrometer; (c) optical path schematics. (Courtesy of Veeco Instruments Inc., Santa
Barbara, CA.)
as versatile and modular as possible is of the utmost

importance. The ability to incorporate additional standard
optical components and easy system reconfiguration
should be first in the designers mind. At the same time,
once a system is set up for operation, it should allow for
easy, straightforward, simple operation, and robustness,
which are sometimes contradictory requirements. The
design of the NSOM head, when having the spectroscopy
package in mind, must incorporate extremely accurate
closed-loop scan linearization. While shear-force feedback
keeps superb control of sample-tip distance, the closed-loop
XY scanning is necessary in order to keep the probe at the
selected spatial position during sample collection.
The near-field spectroscopy package is an even more
promising breakthrough technology for life sciences than
for NSOM imaging alone. For the first time, it will allow
qualitative identification of the composition of an observed
sample, via optical spectroscopies, at the spatial resolution
of scanning probe microscopy. The current commercial
systems can distinguish differences in chemical composition of the samples, at the spatial resolution of 50 nm or
better. For combined near-field spectroscopy and imaging,
the signal may be split into the dual-beam path, wherein
one path is used for the imaging detector and another
one for spectroscopy. Such dual-beam solution minimizes
removal, replacement, and realignment of components
when a different mode is desired. The benefit is the ease
of use and robust, reliable operation for separate or simultaneous transmission and reflection measurements.
In the example of the Aurora-3 optical path system, the
light from both objectives is redirected out the side ports by
two front-surface mirrors, which are reflection-coated for
optimal visible/near-IR (NIR) operation. The near-field
transmission and reflection light is collected in the far field
by precisely aligned microscope objectives to provide high
quality collimated (parallel) beams, with a nominal 7 mm
diameter. This system design allows for NSOM spectroscopic operation with standard one-half in. diameter optics,
though for ease of alignment and handling, 1 in. diameter
optics is generally recommended. Furthermore, the reflection path is carefully aligned to match the transmission
path. Such integrated solutions minimize removal, replacement, and realignment of components when a different
mode is desired. The integrated reflection path is always
available for use with any sample and never requires
removal or realignment in the microscope with normal
use. As there are many variations in the spectroscopy
setup, it is important for an NSOM system that is
intended for spectroscopic use to be capable of utilizing
the standard optical element, so users can customize the
system using standard optical poles and optical bench
mounting systems.
Evaluation
While the first NSOM was invented just 2 years after the
AFM, its widespread use has just started to pick up in the
last several years. The reasons for this lag are severalfold:
the first NSOM images were hard to interpret, and as they
were achieved without topography modulation, the contrast in the images was not intuitive; there were no com-
445
mercial instruments available until the mid-1990s, thus all

users needed to build their own systems; the use of the
home-built and first commercial instrument and its alignment procedures were cumbersome and complicated for
any user who was not skilled in optoelectronics development; and the resolution of the systems depended on each
individual sample. However, the field is changing rapidly,
and with introduction of the latest, third generation of
commercial systems (1316), such as the Aurora-3 from
Veeco Instruments Inc., MultiView 400 from Nanonics
Imaging Ltd, Smena from NT-MDT, and AlphaSNOM from
WiTec GmbH, the ease of use is comparable with the
standard scanning probe microscope and is within the skill
set of average life sciences user. Furthermore, with
improvement in the serial production of the probes and
quality control in recent years, the resolution of the NSOM
system is becoming more uniform, and resolution expectations can be met with most samples.
The way to evaluate and measure resolution of the
NSOM instrument is by utilizing Fishers projection masks
(35). It is virtually accepted as a standard for evaluating
NSOM resolution and quality of image topographic modulation, the latter one by comparing the topographic with
the optical image. The Fisher project mask is a regular
hexagonal array of metallic spikes (Fig. 9). It is produced by
having the monolayer of the monodispersed polymer
spheres coated with metallic coating, and the spheres
subsequently dissolved with organic solvents. What is left
is the regular, closed-packed hexagonal matrix of metallic
spikes. In transmission mode, the spike is seen as a dark
spot on the optical micrography, while in reflection mode,
the spike is a bright spot, and the void space is dark.
The near-field optical imaging obviously provides two
great advantages over other types of imaging: its ability to
simultaneously acquire topography, in a scanning-force
manner, and an optical image. The optical image carries
a plethora of different information that can be furthermore
extracted. The most important advantage is that there are
many different ways to extract direct or indirect information on spatial distribution of chemical composition of the
observed sample, even on the single molecular level.
Furthermore, at current state-of-the-art commercial
NSOM instrumentation, the near-field imaging and spectroscopy can be performed at a resolution at least four
times as high as the resolution of the best optical confocal
microscopes. The obvious applications in the biomedical
field are all of the applications for which the standard
confocal and inverted fluorescence microscope is used
today, but at the same time, done at much higher resolution (3640). Examples of life sciences and biomedical
applications involve optical ultramicroscopy of cells; optical
imaging of cell organelles; imaging and spectroscopy of
individual molecules and macromolecules; in vivo tracking
of molecular events and endocytosis; and high resolution
chromosome labeling [i.e., high resolution fluorescence
in situ hybridization (FISH) (39) applications]. Some of these
applications, like molecular tracking, and high resolution
FISH are unachievable with other methods.
The molecular tracking applications are based on fluorescence, and in the future, Raman SERS applications will
revolutionize our way of understanding how cellular
446
Figure 9. Fisher mask, typical structure for NSOM evaluation and calibration: (a) topographic image
produced with shear-force feedback; (b) NSOM transmission mode image of the same area. The images
were produced on the Aurora-3 NSOM, (Courtesy of Veeco Instruments Inc., Santa Barbara, CA.)
mechanisms work, and will bring significant contributions

to practical pharmacological applications, such as drug
candidates, where their target will be able to be imaged,
measured, and tracked during action. Additionally, this
application will add considerably to the body of knowledge
of macromolecular interactions and in the mapping of the
interactome of proteins, enzymes, and nucleic acids within
the cell. The simplest way will be by expressing differently
colored fluorescence proteins, fused with ligand and target,
and tracking their spatial positions within the cell, and
from the fluorescence resonance energy transfer or fluorescence intermittency, following their interactions. No
other method is capable of achieving the resolution necessary to understand molecular machines in vivo. Another
futuristic application of NSOM is in the ultrahigh density genomics and proteomics array, which theoretically
can be packed at the density range higher than the
wavelengths.
Some of the more futuristic life science applications
will come with the integration of NSOM with mass
spectrometry. Zenobis group (25) demonstrated that
the apertured NSOM is capable of doing the nanoablation
of structures and thus can feed the mass spectrometer
with material ablated with the resolution of a few tens of
nanometer. This can dramatically improve the knowledge
of spatial locations of proteins, and proteinprotein complexes and interactions.
The NSOM-based FISH (38) can have a large impact on
the future of molecular in vitro diagnostics because it will
allow FISH to be applied on the shorter segments of DNA.
This will permit many additional applications by painting
shorter genes of this simple, chromosome painting technique, which are unachievable with far-field fluorescence or
confocal-fluorescence equipment. If designed in a simple
and high throughput manner, this may become a standard
diagnostic instrument for molecular cytogenetics diagnostics in pathology labs.
It is expected that the next, fourth-generation instrumentation, currently in the design stage, will allow more
routine imaging with a level of technical expertise, which is
necessary for practical applications comparable with running todays confocal microscope.
Examples of the NSOMs biological applications are
presented in Fig. 10ac. Figure 10a is an example of the
protein localization imaging beyond diffraction limit. In
this particular image, the fibroblast cells were labeled with
green fluorescence protein (GFP). The image on the left
represents the shear-force micrography, which corresponds to the topographic image, while the image on the
right is the GFP fluorescence image. Spatial distribution of
the GFP can easily be observed within the cell at a resolution far exceeding the diffraction limits. This technique can
be used to track the protein synthesis and trafficking
within the cell if the targeted protein is fused with the
fluorescence label, such as GFP or YFP. Another unique
NSOM application is in optical characterization of the supramolecular and macromolecular assemblies. Figure 10b
represents topographic, shear force (left) and NSOM transmission image (right) of the interband region of a polytene
chromosome. In the optical image, the chromatin matter can
be distinguished from the DNA based on the optical contrast,
which is not possible based on the pure topography. Finally,
Fig. 10c represents far-field optical transmission image of
slice of the muscle tissue (left) and shear-force topography
and near-field optical image, on the right top and bottom,
respectively. The near-field imaging reveals the fine structure of the muscular fiber, its cell membrane, myofibrils, and
endoplasmatic reticulum structure, in a similar manner as
using the transmission electron microscopy.
Besides imaging, the near-field optical microscopy-like
setup has promise for use in the nanoscale lithography (41),
and for high density data storage (42). Nanoscale lithography will have applications in the preparation of tissue
growth matrix and scaffolds, especially for the growth of
neurons, while the high voluminous data storage will have
a plethora of medical applications for storing ever-growing
informational content of both imaging and high throughput diagnostics data.
In conclusion, near-field optical imaging is still a developing technique that shows much promise for biomedical
447
Figure 10. Examples of biomedical

and life-sciences applications of NSOM
imaging: (a) Shear-force topography and
near-field fluorescence from GFPlabeled fibroblast cells. Images of the
fibrorblast cells are prepared by
growing them directly on glass cover
slides and subsequent labeling. They
are imaged in air at a scan speed of 1
Hz, no photodegradation was observed
throughout
the
measurement.
(Courtesy of Renato Zenobi, ETH,
Zurich, # Renato Zenobi and ETH
Zurich, Switzerland). (b) Shear force
and NSOM image of the interband
region of a polytene chromosome.
(Courtesy of Sid Ragona and Phil
Haydon, Laboratory of Cellular
Signaling, Dept. of Zoology and
Genetics, Iowa State University, Ames,
IA, and Veeco Instruments Inc.). (c) Farfield, differential contrast optical
microscopy of the muscle tissue, and
details of the muscle cell by shear force
andNSOM.(CourtesyofSidRagonaand
Phil Haydon, Laboratory of Cellular
Signaling, Department of Zoology and
Genetics, Iowa State University, Ames,
IA and Veeco Instruments Inc.)
448
applications. This review presents the state-of-the-art NSOM

technology up to the second quarter of 2005. It is certain that
by the time of the next edition of this Encyclopedia, there
will be many other technological and advancements in the
field of biomedical applications of near-field optics.
BIBLIOGRAPHY
1. Abbe E. Beitrsge zur theorie des mikroskops und mikroskopichen wahrnehmung. Arch Mikroskop Anat 1873;9:413.
2. Synge EH. A suggested method for extending the microscopic
resolution into the ultramicroscopic region. Phil Mag 1928;6:
356362.
3. Synge EH. An application of piezoelectricity to microscopy.
Phil Mag 1932;13:297300.
4. OKeefe JA. Resolving power of visible light. J Opt Soc Am
1956;46:359362.
5. Ash A, Nichols G. Super-resolution aperture scanning microscope. Nature London 1972;237:510511.
6. Pohl DW, Denk W, Lanz M. Optical stethoscopyImage recording with resolution lambda/20. Appl Phys Lett 1984;44: 651653.
7. Lewis A, Isaacson M, Harootunian A, Murray A. Development of 500A spatial-resolution light-microscope. 1. Light is
efficiently transmitted through gamma-16 diameter apertures. Ultramicroscopy 1984;13:227231.
8. Rasmussen A, Deckert V. New dimension in nano-imaging:
breaking through the diffraction limit with scanning near-field
optical microscopy. Anal Bioanal Chem 2005;381:165172.
9. Edidin M. Near-field scanning optical microscopy, a siren call
to biology. Traffic 2001;2:797803.
10. Dunn RC. Near-field scanning optical microscopy. Chem Rev
1999;99:28912899.
11. Micic M. Near-field scanning optical microscopy and spectroscopy advance. Photonics Spectra 2005;38:124125.
12. De Serio M, Zenobi R, Deckert V. Looking at the nanoscale:
scanning near-field optical microscopy. TRAC-Trends Anal
Chem 2003;22:7077.
13. Product info: Aurora-3 NSOM , Veeco Instruments Inc.,
Santa Barbara, CA. Available at http://www.veeco.com.
14. Product info: Multiview Series, Nanonics Imaging Ltd., Jerusalem, Israel. Available at http://www.nanonics.co.il.
15. Product info: Alpha SNOM, Witec Wissenschaftliche Instrumente und Technologie. Available at GmbH GmbH,Ulm,
Germany, http://www.witec.de.
16. Product info: Solver SNOM, NT-MDT Co, Zelenogorod, Russia. Available at http://www.ntmdt.ru.
17. Paesler MA, Moyer P. Near-Field Optics Theory, Instrumentation and Applications. New York: John Wiley & Sons; 1996.
18. Curie P, Curie J. Developement, par pression de lelectricite
polaire dans les cristaux hemiedres a faces inclinees. Comp
Ren 1880;91:291295.
19. Spanner K. Micro Positioning, Nano Positioning, Nano Automation: Solution for Cutting Edge Technologies (product
catalog), Karslruhe, Physik Instrumente (PI) GmbH & Co
KG. Available at http://www.pi.ws. Accessed 2005.
20. Betzig E, Finn PL, Weiner JS. Combined shear force and
near-field scanning optical microscope. Appl Phys Lett
1992;60: 24842486.
21. Garcia-Parajo M, Tate T, Chen Y. Gold-coated parabolic
tapers for scanning near-field optical microscopy: Fabrication
and optimisation. Ultramicroscopy 1995;61:155163.
22. Pangaribuan T, et al. Reproducible fabrication technique of
nanometric tip diameter fiber probe for photon scanning
tunneling microscope. Jpn J Appl Phys 1992;31:L1302
L1304.
23. Radojewski R, Grabijec P. Combined SNOM/AFM microscopy

with micromachined nanoapertures. Mater SciPoland
2003;21:321332.
24. Takahashi S, Dickson W, Pollard R, Zaytas A. Near-field
magneto-optical analysis in reflection mode SNOM. Ultramicroscopy 2004;100(34):443447.
25. Stockle R, et al. Nanoscale atmospheric pressure laser ablation mass spectrometry. Anal Chem 2001;73:1391402.
26. Chovin A, Garrigue P, Servant L, Sojic N. Electrochemical modulation of remote fluorescence imaging at an ordered opto-electrochemical nanoaperture array. Chemphyschem 2004;5:11251132.
27. Inoye Y, Kawata S. Near-field scanning optical microscope
with a metallic probe tip. Opt Lett 1994;19:159161.
28. Keilmann F, Hillenbrand R. Near-field microscopy by elastic
light scattering from a tip. Philas Trans R Soc Sci A
2004;362:787805.
29. Hu DH, et al. Correlated topographic and spectroscopic imaging beyond diffraction limit by atomic force microscopy
metallic tip enhanced near-field fluorescence lifetime microscopy. Rev Sci Instr 2003;74:33473355.
30. Micic M, Klymyshin N, Suh YD, Lu HP. Finite element
method simulation of the field distribution for AFM tip
enhanced surface enhanced Raman scanning microscopy. J
Phys Chem B 2003;107:15741584.
31. Sun WX, Shen ZX. Near-field scanning Raman microscopy using
apertureless probes. J Raman Spectrosc 2003;34:668676.
32. Richards D. Near-field microscopy: Throwing light on the
nanoworld Philas Trans R Soc Sci A 2003;361:28432857.
33. Vanden Bout DA, Kerimo J, Higgins DA, Barbara PF. Spatially Resolved Spectral Inhomogeneities in Small Molecular
Crystals Studied by Near Field Scanning. Opt Microsc J Phys
Chem 1996;100:1184311850.
34. Puestow R. Configuring Aurora-3 for Spectroscopy, application note, Veeco Instruments Inc, Santa Barbara, CA, 2003.
35. Fischer UC, et al. Latex bead projection nanopatterns. Surf
Interface Anal 2002;33:7580.
36. de Lange F, et al. Cell biology beyond the diffraction limit: Nearfield scanning optical microscopy. J Cell Sci 2001;114: 41534160.
37. Subramaniam V, Kirsch AK, Jovin TM. Cell biological applications of scanning near-field optical microscopy (SNOM).
Cell Molec Biol 1998;44:689700.
38. Lewis A, et al. Near-field scanning optical microscopy in cell
biology. Trends Cell Biol 1999;9:7073.
39. Fukushi D, et al. Scanning near-field optical/atomic force microscopy detection of fluorescence in situ hybridization signals
beyond the optical limit. Exp Cell Res 2003;289:237244.
40. Krishnan RV, Varma R, Mayor S. Fluorescence methods to
probe nanometer-scale organization of molecules in living cell
membranes. J Fluoresc 2001;11:211226.
41. Dryakhulshin VF, Klimov AY, Rogov VV, Vostkov NV. Nearfield optical lithography method for fabrication of nanodimensional objects. Appl Surf Sci 2005;248:200203.
42. Ferri V, et al. Near-field optical addressing of luminescent
photoswitchable supramolecular system embedded in inert
polymer matrices. Nano Lett 2004;4:835859.
Further Reading
Prasad PN. Nanophotonics. New York: John Wiley & Sons; 2004.
Courion D. Near Field Microscopy and Near Field Optics. London:
Imperial College Press; 2003.
Paul DW, Courion D. Near Field Optics. Arc-et Senans: Kulwer
Academic Publisher; 1993.
Taatjes DJ, Brooke MT. Cell Imaging Techniques: Methods and
Protocolos. Totowa: Humana Press; 2005.
See also MICROSCOPY,
CONFOCAL; MICROARRAYS; NANOPARTICLES.
NATHAN S. CLAXTON
THOMAS J. FELLERS
MICHAEL W. DAVIDSON
The Florida State University
Tallahassee, Florida
INTRODUCTION
The technique of laser scanning and spinning disk confocal
fluorescence microscopy has become an essential tool in
biology and the biomedical sciences, as well as in materials
science due to attributes that are not readily available
using other contrast modes with traditional optical microscopy (112). The application of a wide array of new synthetic and naturally occurring fluorochromes has made it
possible to identify cells and submicroscopic cellular components with a high degree of specificity amid nonfluorescing
material (13). In fact, the confocal microscope is often capable of revealing the presence of a single molecule (14).
Through the use of multiply labeled specimens, different
probes can simultaneously identify several target molecules
simultaneously, both in fixed specimens and living cells and
tissues (15). Although both conventional and confocal microscopes cannot provide spatial resolution below the diffraction limit of specific specimen features, the detection of
fluorescing molecules below such limits is readily achieved.
The basic concept of confocal microscopy was originally
developed by Minsky in the mid-1950s (patented in 1961)
when he was a postdoctoral student at Harvard University
(16,17). Minsky wanted to image neural networks in
unstained preparations of brain tissue and was driven
by the desire to image biological events as they occur in
living systems. Minskys invention remained largely unnoticed, due most probably to the lack of intense light sources
necessary for imaging and the computer horsepower
required to handle large amounts of data. Following Minskys work, Egger and Petran (18) fabricated a multiplebeam confocal microscope in the late-1960s that utilized a
spinning (Nipkow) disk for examining unstained brain
sections and ganglion cells. Continuing in this arena,
Egger went on to develop the first mechanically scanned
confocal laser microscope, and published the first recognizable images of cells in 1973 (19). During the late-1970s and
the 1980s, advances in computer and laser technology,
coupled to new algorithms for digital manipulation of
images, led to a growing interest in confocal microscopy (20).
Fortuitously, shortly after Minskys patent had expired,
practical laser-scanning confocal microscope designs were
translated into working instruments by several investigators. Dutch physicist Brakenhoff developed a scanning
confocal microscope in 1979 (21), while almost simultaneously, Sheppard contributed to the technique with a
theory of image formation (22). Wilson, Amos, and White
nurtured the concept and later (during the late-1980s)
demonstrated the utility of confocal imaging in the examination of fluorescent biological specimens (20,23). The first
commercial instruments appeared in 1987. During the
1990s, advances in optics and electronics afforded more
stable and powerful lasers, high efficiency scanning mirror
449
units, high throughput fiber optics, better thin-film dielectric coatings, and detectors having reduced noise characteristics (1). In addition, fluorochromes that were more
carefully matched to laser excitation lines were beginning
to be synthesized (13). Coupled to the rapidly advancing
computer processing speeds, enhanced displays, and largevolume storage technology emerging in the late-1990s, the
stage was set for a virtual explosion in the number of
applications that could be targeted with laser scanning
confocal microscopy.
Modern confocal microscopes can be considered as completely integrated electronic systems where the optical microscope plays a central role in a configuration that consists of
one or more electronic detectors, a computer (for image display, processing, output, and storage), and several laser
systems combined with wavelength selection devices and a
beam scanning assembly. In most cases, integration between
the various components is so thorough that the entire confocal microscope is often collectively referred to as a digital or
video imaging system capable of producing electronic
images (24). These microscopes are now being employed
for routine investigations on molecules, cells, and living
tissues that were not possible just a few years ago (15).
Confocal microscopy offers several advantages over conventional widefield optical microscopy, including the ability to control depth of field, elimination, or reduction of
background information away from the focal plane (that
leads to image degradation), and the capability to collect
serial optical sections from thick specimens. The basic key
to the confocal approach is the use of spatial filtering
techniques to eliminate out-of-focus light or glare in specimens whose thickness exceeds the immediate plane of
focus. There has been a tremendous explosion in the popularity of confocal microscopy in recent years (14,6,7), due
in part to the relative ease with which extremely high
quality images can be obtained from specimens prepared
for conventional fluorescence microscopy, and the growing
number of applications in cell biology that rely on imaging,
both fixed and living cells and tissues. In fact, confocal
technology is proving to be one of the most important
advances ever achieved in optical microscopy.
In a conventional widefield optical epi-fluorescence
microscope, secondary fluorescence emitted by the specimen often occurs through the excited volume and obscures
resolution of features that lie in the objective focal plane
(25). The problem is compounded by thicker specimens (>2
mm), which usually exhibit such a high degree of fluorescence emission that most of the fine detail is lost. Confocal
microscopy provides only a marginal improvement in both
axial (z; parallel to the microscope optical axis) and lateral
(x and y; dimensions in the specimen plane) optical resolution, but is able to exclude secondary fluorescence in areas
removed from the focal plane from resulting images (26
28). Even though resolution is somewhat enhanced with
confocal microscopy over conventional widefield techniques (1), it is still considerably less than that of the transmission electron microscope (TEM). In this regard, confocal
microscopy can be considered a bridge between these two
classical methodologies.
Illustrated in Fig. 1 are a series of images that compare selected viewfields in traditional widefield and laser
450
Figure 1. Comparison of widefield (upper row) and laser scanning confocal fluorescence microscopy
images (lower row). Note the significant amount of signal in the widefield images from fluorescent
structures located outside of the focal plane. (a) and (b) Mouse brain hippocampus thick section
treated with primary antibodies to glial fibrillary acidic protein (GFAP; red), neurofilaments H
(green), and counterstained with Hoechst 33342 (blue) to highlight nuclei. (c) and (d) Thick section of
rat smooth muscle stained with phalloidin conjugated to Alexa Fluor 568 (targeting actin; red),
wheat germ agglutinin conjugated to Oregon Green 488 (glycoproteins; green), and counterstained
with DRAQ5 (nuclei; blue). (e) and (f) Sunflower pollen grain tetrad autofluorescence.
scanning confocal fluorescence microscopy. A thick (16

mm) section of fluorescently stained mouse hippocampus
in widefield fluorescence exhibits a large amount of glare
from fluorescent structures located above and below the
focal plane (Fig. 1a). When imaged with a laser scanning
confocal microscope (Fig. 1b), the brain thick section
reveals a significant degree of structural detail. Likewise,
widefield fluorescence imaging of rat smooth muscle fibers
stained with a combination of Alexa Fluor dyes produce
blurred images (Fig. 1c) lacking in detail, while the same
specimen field (Fig. 1d) reveals a highly striated topography when viewed as an optical section with confocal
microscopy. Autofluorescence in a sunflower (Helianthus
annuus) pollen grain tetrad produces a similar indistinct
outline of the basic external morphology (Fig. 1e), but
yields no indication of the internal structure in widefield
mode. In contrast, a thin optical section of the same grain
(Fig. 1f) acquired with confocal techniques displays a
dramatic difference between the particle core and the
surrounding envelope. Collectively, the image comparisons
in Fig. 1 dramatically depict the advantages of achieving
very thin optical sections in confocal microscopy. The ability
of this technique to eliminate fluorescence emission from
regions removed from the focal plane offsets it from traditional forms of fluorescence microscopy.
PRINCIPLES OF CONFOCAL MICROSCOPY

The confocal principle in epi-fluorescence laser scanning
microscope is diagrammatically presented in Fig. 2. Coherent light emitted by the laser system (excitation source)
passes through a pinhole aperture that is situated in a
conjugate plane (confocal) with a scanning point on the
specimen and a second pinhole aperture positioned in front
of the detector (a photomultiplier tube). As the laser is
reflected by a dichromatic mirror, and scanned across the
specimen in a defined focal plane, secondary fluorescence
emitted from points on the specimen (in the same focal
plane) pass back through the dichromatic mirror, and are
focused as a confocal point at the detector pinhole aperture.
The significant amount of fluorescence emission that
occurs at points above and below the objective focal plane is
not confocal with the pinhole (termed out-of-focus light
rays in Fig. 2) and forms extended Airy disks in the
aperture plane (29). Because only a small fraction of the
out-of-focus fluorescence emission is delivered through
the pinhole aperture, most of this extraneous light is not
detected by the photomultiplier and does not contribute to
the resulting image. The dichromatic mirror, barrier filter,
and excitation filter perform similar functions to identical
components in a widefield epi-fluorescence microscope (30).
Figure 2. Schematic diagram of the optical pathway and

principal components in a laser scanning confocal microscope.
Refocusing the objective in a confocal microscope shifts the

excitation and emission points on a specimen to a new
plane that becomes confocal with the pinhole apertures of
the light source and detector.
In traditional widefield epi-fluorescence microscopy, the
entire specimen is subjected to intense illumination from
an incoherent mercury or xenon arc-discharge lamp, and
the resulting image of secondary fluorescence emission can
be viewed directly in the eyepieces or projected onto the
surface of an electronic array detector or traditional film
plane. In contrast to this simple concept, the mechanism of
image formation in a confocal microscope is fundamentally
different (31). As discussed above, the confocal fluorescence
microscope consists of multiple laser excitation sources, a
scan head with optical and electronic components, electronic detectors (usually photomultipliers), and a computer
for acquisition, processing, analysis, and display of images.
The scan head is at the heart of the confocal system and
is responsible for rasterizing the excitation scans, as well as
collecting the photon signals from the specimen that are
required to assemble the final image (1,57). A typical scan
head contains inputs from the external laser sources,
fluorescence filter sets and dichromatic mirrors, a galvanometer-based raster scanning mirror system, variable pinhole apertures for generating the confocal image, and
photomultiplier tube detectors tuned for different fluorescence wavelengths. Many modern instruments include
diffraction gratings or prisms coupled with slits positioned
near the photomultipliers to enable spectral imaging
(also referred to as emission fingerprinting) followed
by linear unmixing of emission profiles in specimens
labeled with combinations of fluorescent proteins or fluorophores having overlapping spectra (3238). The general
arrangement of scan head components is presented in
Fig. 3 for a typical commercial unit.
451
Figure 3. Three-channel spectral imaging laser scanning

microscope confocal scan head with SIM scanner laser port. The
SIM laser enables simultaneous excitation and imaging of the
specimen for photobleaching or photoactivation experiments. Also
illustrated are ports for a visible, ultraviolet (UV), and infrared (IR)
laser, as well as an arc discharge lamp port for widefield observation.
In epi-illumination scanning confocal microscopy, the

laser light source and photomultiplier detectors are both
separated from the specimen by the objective, which functions as a well-corrected condenser and objective combination. Internal fluorescence filter components (e.g., the
excitation and barrier filters and the dichromatic mirrors)
and neutral density filters are contained within the scanning unit (see Fig. 3). Interference and neutral density
filters are housed in rotating turrets or sliders that can be
inserted into the light path by the operator. The excitation
laser beam is connected to the scan unit with a fiber optic
coupler followed by a beam expander that enables the thin
laser beam wrist to completely fill the objective rear aperture (a critical requirement in confocal microscopy).
Expanded laser light that passes through the microscope
objective forms an intense diffraction-limited spot that is
scanned by the coupled galvanometer mirrors in a raster
pattern across the specimen plane (point scanning).
One of the most important components of the scanning
unit is the pinhole aperture, which acts as a spatial filter at
the conjugate image plane positioned directly in front of the
photomultiplier (39). Several apertures of varying diameter are usually contained on a rotating turret that
enables the operator to adjust pinhole size (and optical
section thickness). Secondary fluorescence collected by the
objective is descanned by the same galvanometer mirrors
that form the raster pattern, and then passes through a
barrier filter before reaching the pinhole aperture (40).
The aperture serves to exclude fluorescence signals from
out-of-focus features positioned above and below the focal
plane, which are instead projected onto the aperture as
Airy disks having a diameter much larger than those
forming the image. These oversized disks are spread over
452
Figure 4. Widefield versus confocal microscopy illumination

volumes, demonstrating the difference in size between point
scanning and widefield excitation light beams.
a comparatively large area so that only a small fraction of

light originating in planes away from the focal point passes
through the aperture. The pinhole aperture also serves to
eliminate much of the stray light passing through the optical
system. Coupling of aperture-limited point scanning to a
pinhole spatial filter at the conjugate image plane is an
essential feature of the confocal microscope.
When contrasting the similarities and differences
between widefield and confocal microscopes, it is often
useful to compare the character and geometry of specimen
illumination utilized for each of the techniques. Traditional
widefield epi-fluorescence microscope objectives focus
a wide cone of illumination over a large volume of the
specimen (41), which is uniformly and simultaneously
illuminated (as illustrated in Fig. 4a). A majority of the
fluorescence emission directed back toward the microscope
is gathered by the objective (depending on the numerical
aperture) and projected into the eyepieces or detector. The
result is a significant amount of signal due to emitted
background light and autofluorescence originating from
areas above and below the focal plane, which seriously
reduces resolution and image contrast.
The laser illumination source in confocal microscopy is
first expanded to fill the objective rear aperture, and then
focused by the lens system to a very small spot at the focal
plane (Fig. 4b). The size of the illumination point ranges
from 0.25 to 0.8 mm in diameter (depending on the
objective numerical aperture) and 0.5 to 1.5 mm deep at
the brightest intensity. Confocal spot size is determined by
the microscope design, wavelength of incident laser light,
objective characteristics, scanning unit settings, and the
specimen (41). Figure 4 presents a comparison between the
typical illumination cones of a widefield (Fig. 4a) and point
scanning confocal (Fig. 4b) microscope at the same numerical aperture. The entire depth of the specimen over a wide
area is illuminated by the widefield microscope, while the
sample is scanned with a finely focused spot of illumination
that is centered in the focal plane in the confocal microscope.
In laser scanning confocal microscopy, the image of an
extended specimen is generated by scanning the focused
beam across a defined area in a raster pattern controlled by
two high speed oscillating mirrors driven with galvanometer motors. One of the mirrors moves the beam from
left to right along the x lateral axis, while the other
translates the beam in the y direction. After each single
scan along the x axis, the beam is rapidly transported back
to the starting point and shifted along the y axis to begin a
new scan in a process termed flyback (42). During the
flyback operation, image information is not collected. In
this manner, the area of interest on the specimen in a

single focal plane is excited by laser illumination from the
scanning unit.
As each scan line passes along the specimen in the lateral
focal plane, fluorescence emission is collected by the objective and passed back through the confocal optical system.
The speed of the scanning mirrors is very slow relative to the
speed of light, so the secondary emission follows a light path
along the optical axis that is identical to the original excitation beam. Return of fluorescence emission through the
galvanometer mirror system is referred to as descanning
(40,42). After leaving the scanning mirrors, the fluorescence
emission passes directly through the dichromatic mirror
and is focused at the detector pinhole aperture. Unlike
the raster scanning pattern of excitation light passing over
the specimen, fluorescence emission remains in a steady
position at the pinhole aperture, but fluctuates with respect
to intensity over time as the illumination spot traverses the
specimen producing variations in excitation.
Fluorescence emission that is passed through the pinhole aperture is converted into an analog electrical signal
having a continuously varying voltage (corresponding to
intensity) by the photomultiplier. The analog signal is
periodically sampled and converted into pixels by an
analog-to-digital (A/D) converter housed in the scanning
unit or the accompanying electronics cabinet. The image
information is temporarily stored in an image frame buffer
card in the computer and displayed on the monitor. Note
that the confocal image of a specimen is reconstructed,
point by point, from emission photon signals by the photomultiplier and accompanying electronics, yet never exists
as a real image that can be observed through the microscope eyepieces.
LASER SCANNING CONFOCAL MICROSCOPE

CONFIGURATION
Basic microscope optical system characteristics have
remained fundamentally unchanged for many decades
due to engineering restrictions on objective design, the
static properties of most specimens, and the fact that
resolution is governed by the wavelength of light (110).
However, fluorescent probes that are employed to add
contrast to biological specimens and, and other technologies associated with optical microscopy techniques, have
improved significantly. The explosive growth and development of the confocal approach is a direct result of a renaissance in optical microscopy that has been largely fueled by
advances in modern optical and electronics technology.
Among these are stable multiwavelength laser systems
that provide better coverage of the uv, visible, and nearIR spectral regions, improved interference filters (including dichromatic mirrors, barrier, and excitation filters),
sensitive low noise wide-band detectors, and far more
powerful computers. The latter are now available with
relatively low cost memory arrays, image analysis software
packages, high resolution video displays, and high quality
digital image printers. The flow of information through a
modern confocal microscope is presented diagrammatically in Fig. 5 (2).
453
computer, which also controls the scanning mirrors and/or

other devices to facilitate the collection and display of
images. After a series of images (usually serial optical
sections) has been acquired and stored on digital media,
analysis can be conducted utilizing numerous image processing software packages available on the host or a secondary computer.
ADVANTAGES AND DISADVANTAGES OF CONFOCAL

MICROSCOPY
Figure 5. Confocal microscope configuration and information

flow schematic diagram.
Although many of these technologies have been developed independently for a variety of specifically targeted
applications, they have been incorporated gradually into
mainstream commercial confocal microscopy systems. In
current microscope systems, classification of designs is
based on the technology utilized to scan specimens (7).
Scanning can be accomplished either by translating the
stage in the x, y, and z directions while the laser illumination spot is held in a fixed position, or the beam itself can
be raster-scanned across the specimen. Because threedimensional (3D) translation of the stage is cumbersome
and prone to vibration, most modern instruments employ
some type of beam-scanning mechanism.
In modern confocal microscopes, two fundamentally
different techniques for beam scanning have been developed. Single-beam scanning, one of the more popular
methods employed in a majority of the commercial laser
scanning microscopes (43), uses a pair of computer-controlled galvanometer mirrors to scan the specimen in a
raster pattern at a rate of approximately one frame per
second. Faster scanning rates (to near video speed) can be
achieved using acoustooptic devices or oscillating mirrors.
In contrast, multiple-beam scanning confocal microscopes
are equipped with a spinning Nipkow disk containing an
array of pinholes and microlenses (4446). These instruments often use arc-discharge lamps for illumination
instead of lasers to reduce specimen damage and enhance
the detection of low fluorescence levels during real-time
image collection. Another important feature of the multiplebeam microscopes is their ability to readily capture images
with an array detector, such as a charge-coupled device
(CCD) camera system (47).
All modern laser scanning confocal microscope designs
are centered on a conventional upright or inverted
research level optical microscope. However, instead of
the standard tungstenhalogen or mercury (xenon) arcdischarge lamp, one or more laser systems are used as a
light source to excite fluorophores in the specimen. Image
information is gathered point by point with a specialized
detector, such as a photomultiplier tube or avalanche
photodiode, and then digitized for processing by the host
The primary advantage of laser scanning confocal microscopy is the ability to serially produce thin (0.51.5 mm)
optical sections through fluorescent specimens that have a
thickness ranging up to 50 mm or more (48). The image
series is collected by coordinating incremental changes in
the microscope fine focus mechanism (using a stepper
motor) with sequential image acquisition at each step.
Image information is restricted to a well-defined plane,
rather than being complicated by signals arising from
remote locations in the specimen. Contrast and definition
are dramatically improved over widefield techniques due to
the reduction in background fluorescence and improved
signal to noise (48). Furthermore, optical sectioning eliminates artifacts that occur during physical sectioning and
fluorescent staining of tissue specimens for traditional
forms of microscopy. The noninvasive confocal optical sectioning technique enables the examination of both living
and fixed specimens under a variety of conditions with
enhanced clarity.
With most confocal microscopy software packages, optical sections are not restricted to the perpendicular lateral
(xy) plane, but can also be collected and displayed in
transverse planes (1,58,49). Vertical sections in the xz
and yz planes (parallel to the microscope optical axis) can
be readily generated by most confocal software programs.
Thus, the specimen appears as if it had been sectioned in a
plane that is perpendicular to the lateral axis. In practice,
vertical sections are obtained by combining a series of xy
scans taken along the z axis with the software, and then
projecting a view of fluorescence intensity as it would
appear should the microscope hardware have been capable
of physically performing a vertical section.
A typical stack of optical sections (often termed a z
series) through a Lodgepole Pine tree pollen grain revealing internal variations in autofluorescence emission wavelengths is illustrated in Fig. 6. Optical sections were
gathered in 1.0 mm steps perpendicular to the z axis
(microscope optical axis) using a laser combiner featuring
an argon ion (488 nm; green fluorescence), a green helium
neon (543 nm; red fluorescence), and a red heliumneon
(633 nm; fluorescence pseudocolored blue) laser system.
Pollen grains from this and many other species range
between 10 and 40 mm in diameter and often yield blurred
images in wide-field fluorescence microscopy (see Fig. 1c),
which lack information about internal structural details.
Although only 12 of the >36 images collected through this
series are presented in the figure, they represent individual focal planes separated by a distance of 3 mm and
provide a good indication of the internal grain structure.
454
Figure 6. Lodgepole pine (Pinus contorta) pollen grain optical

sections. Bulk pollen was mounted in CytoSeal 60 and imaged with
a 100 oil immersion objective (no zoom) in 1 mm axial steps. Each
image in the sequence (112) represents the view obtained from
steps of 3 mm.
In specimens more complex than a pollen grain, complex

interconnected structural elements can be difficult to discern from a large series of optical sections sequentially
acquired through the volume of a specimen with a laser
scanning confocal microscope. However, once an adequate
series of optical sections has been gathered, it can be further
processed into a 3D representation of the specimen using
volume-rendering computational techniques (5053). This
approach is now in common use to help elucidate the
numerous interrelationships between structure and function of cells and tissues in biological investigations (54). In
order to ensure that adequate data is collected to produce a
representative volume image, the optical sections should be
recorded at the appropriate axial (z step) intervals so that
the actual depth of the specimen is reflected in the image.
Most of the software packages accompanying commercial confocal instruments are capable of generating composite and multidimensional views of optical section data
acquired from z-series image stacks. The 3D software
packages can be employed to create either a single 3D
representation of the specimen (Fig. 7) or a video (movie)
sequence compiled from different views of the specimen
volume. These sequences often mimic the effect of rotation
or similar spatial transformation that enhances the appreciation of the specimens 3D character. In addition, many
software packages enable investigators to conduct measurements of length, volume, and depth, and specific parameters
of the images, such as opacity, can be interactively altered to
reveal internal structures of interest at differing levels
within the specimen (54).
Typical 3D representations of several specimens examined by serial optical sectioning are presented in Fig. 7. A
series of sunflower pollen grain optical sections was combined to produce a realistic view of the exterior surface
(Fig. 7a) as it might appear if being examined by a scanning
electron microscope (SEM). The algorithm utilized to
construct the 3D model enables the user to rotate the
Figure 7. Three-dimensional volume renders from confocal

microscopy optical sections. (a) Autofluorescence in a series of
sunflower pollen grain optical sections was combined to produce
a realistic view of the exterior surface. (b) Mouse lung tissue thick
(16 mm) section. (c) Rat brain thick section. These specimens were each
labeled with several fluorophores (blue, green, and red fluorescence)
and the volume renders were created from a stack of 3045 optical
sections. (d) Autofluorescence in a thin section of fern root.
pollen grain through 3608 for examination. Similarly, thick

sections (16 mm) of lung tissue and rat brain are presented in
Fig. 7b and 7c, respectively. These specimens were each
labeled with several fluorophores (blue, green, and red
fluorescence) and created from a stack of 3045 optical
sections. Autofluorescence in plant tissue was utilized to
produce the model illustrated in Fig. 7d of a fern root section.
In many cases, a composite or projection view produced from a series of optical sections provides important
information about a 3D specimen than a multidimensional
view (54). For example, a fluorescently labeled neuron
having numerous thin, extended processes in a tissue
section is difficult (if not impossible) to image using
wide-field techniques due to out-of-focus blur. Confocal
thin sections of the same neuron each reveal portions of
several extensions, but these usually appear as fragmented
streaks and dots and lack continuity (53). Composite views
created by flattening a series of optical sections from the
neuron will reveal all of the extended processes in sharp
focus with well-defined continuity. Structural and functional analysis of other cell and tissue sections also benefits
from composite views as opposed to, or coupled with, 3D
volume rendering techniques.
Advances in confocal microscopy have made possible
multidimensional views (54) of living cells and tissues that
include image information in the x, y, and z dimensions as
a function of time and presented in multiple colors (using
two or more fluorophores). After volume processing of
individual image stacks, the resulting data can be displayed as 3D multicolor video sequences in real time. Note
that unlike conventional widefield microscopy, all fluorochromes in multiply labeled specimens appear in register
using the confocal microscope. Temporal data can be collected either from time-lapse experiments conducted over
extended periods or through real-time image acquisition in
smaller frames for short periods of time. The potential for
using multidimensional confocal microscopy as a powerful
tool in cellular biology is continuing to grow as new laser
systems are developed to limit cell damage and computer
processing speeds and storage capacity improves.
Additional advantages of scanning confocal microscopy
include the ability to adjust magnification electronically by
varying the area scanned by the laser without having to
change objectives. This feature is termed the zoom factor,
and is usually employed to adjust the image spatial resolution by altering the scanning laser sampling period
(1,2,8,40,55). Increasing the zoom factor reduces the specimen area scanned and simultaneously reduces the scanning
rate. The result is an increased number of samples along a
comparable length (55), which increases both the image
spatial resolution and display magnification on the host
computer monitor. Confocal zoom is typically employed to
match digital image resolution (8,40,55) with the optical
resolution of the microscope when low numerical aperture
and magnification objectives are being used to collect data.
Digitization of the sequential analog image data collected by the confocal microscope photomultiplier (or similar
detector) facilitates computer image processing algorithms
by transforming the continuous voltage stream into discrete
digital increments that correspond to variations in light
intensity. In addition to the benefits and speed that accrue
from processing digital data, images can be readily prepared
for print output or publication. In carefully controlled
experiments, quantitative measurements of spatial fluorescence intensity (either statically or as a function of time) can
also be obtained from the digital data.
Disadvantages of confocal microscopy are limited primarily to the limited number of excitation wavelengths
available with common lasers (referred to as laser lines),
which occur over very narrow bands and are expensive to
produce in the UV region (56). In contrast, conventional
widefield microscopes use mercury- or xenon-based arcdischarge lamps to provide a full range of excitation wavelengths in the UV, visible, and near-IR spectral regions.
Another downside is the harmful nature (57) of high
intensity laser irradiation to living cells and tissues, an
issue that has recently been addressed by multiphoton and
Nipkow disk confocal imaging. Finally, the high cost of
purchasing and operating multiuser confocal microscope
systems (58), which can range up to an order of magnitude
higher than comparable widefield microscopes, often limits
their implementation in smaller laboratories. This problem
can be easily overcome by cost-shared microscope systems
that service one or more departments in a core facility. The
recent introduction of personal confocal systems has competitively driven down the price of low end confocal microscopes and increased the number of individual users.
CONFOCAL MICROSCOPE LIGHT DETECTORS

In modern widefield fluorescence and laser scanning confocal optical microscopy, the collection and measurement of
455
secondary emission gathered by the objective can be accomplished by several classes of photosensitive detectors (59),
including photomultipliers, photodiodes, and solid-state
CCDs. In confocal microscopy, fluorescence emission is
directed through a pinhole aperture positioned near the
image plane to exclude light from fluorescent structures
located away from the objective focal plane, thus reducing
the amount of light available for image formation, as
discussed above. As a result, the exceedingly low light
levels most often encountered in confocal microscopy necessitate the use of highly sensitive photon detectors that do
not require spatial discrimination, but instead respond
very quickly with a high level of sensitivity to a continuous
flux of varying light intensity.
Photomultipliers, which contain a photosensitive surface that captures incident photons and produces a stream
of photoelectrons to generate an amplified electric charge,
are the popular detector choice in many commercial confocal microscopes (5961). These detectors contain a critical element, termed a photocathode, capable of emitting
electrons through the photoelectric effect (the energy of an
absorbed photon is transferred to an electron) when
exposed to a photon flux. The general anatomy of a photomultiplier consists of a classical vacuum tube in which a
glass or quartz window encases the photocathode and a
chain of electron multipliers, known as dynodes, followed
by an anode to complete the electrical circuit (62). When
the photomultiplier is operating, current flowing between
the anode and ground (zero potential) is directly proportional to the photoelectron flux generated by the photocathode when it is exposed to incident photon radiation.
In a majority of commercial confocal microscopes, the
photomultiplier is located within the scan head or an
external housing, and the gain, offset, and dynode voltage
are controlled by the computer software interface to the
detector power supply and supporting electronics (7). The
voltage setting is used to regulate the overall sensitivity of
the photomultiplier, and can be adjusted independently of
the gain and offset values. The latter two controls are
utilized to adjust the image intensity values to ensure that
the maximum number of gray levels is included in the
output signal of the photomultiplier. Offset adds a positive
or negative voltage to the output signal, and should be
adjusted so that the lowest signals are near the photomultiplier detection threshold (40). The gain circuit multiplies
the output voltage by a constant factor so that the maximum signal values can be stretched to a point just below
saturation. In practice, offset should be applied first before
adjusting the photomultiplier gain (8,40). After the signal
has been processed by the analog-to-digital converter, it is
stored in a frame buffer and ultimately displayed on the
monitor in a series of gray levels ranging from black (no
signal) to white (saturation). Photomultipliers with a
dynamic range of 10 or 12 bits are capable of displaying
1024 or 4096 gray levels, respectively. Accompanying
image files also have the same number of gray levels.
However, the photomultipliers used in a majority of the
commercial confocal microscopes have a dynamic range
limited to 8 bits or 256 gray levels, which in most cases,
is adequate for handling the typical number of photons
scanned per pixel (63).
456
Changes to the photomultiplier gain and offset levels

should not be confused with postacquisition image processing to adjust the levels, brightness, or contrast in the final
image. Digital image processing techniques can stretch
existing pixel values to fill the black-to-white display
range, but cannot create new gray levels (40). As a result,
when a digital image captured with only 200 out of a
possible 4096 gray levels is stretched to fill the histogram
(from black to white), the resulting processed image
appears grainy. In routine operation of the confocal microscope, the primary goal is to fill as many of the gray levels
during image acquisition and not during the processing
stages.
The offset control is used to adjust the background level
to a position near 0 V (black) by adding a positive or
negative voltage to the signal. This ensures that dark
features in the image are very close to the black level of
the host computer monitor. Offset changes the amplitude of
the entire voltage signal, but since it is added to or subtracted from the total signal, it does not alter the voltage
differential between the high and low voltage amplitudes
in the original signal. For example, with a signal ranging
from 4 to 18 V that is modified with an offset setting of 4 V,
the resulting signal spans 014 V, but the difference
remains 14 V.
Figure 8 presents a series of diagrammatic schematics
of the unprocessed and adjusted output signal from a
photomultiplier and the accompanying images captured
with a confocal microscope of a living adherent culture of
Indian Muntjac deer skin fibroblast cells treated with
MitoTracker Red CMXRos, which localizes specifically in
the mitochondria. Figure 8a illustrates the raw confocal
image along with the signal from the photomultiplier. After
applying a negative offset voltage to the photomultiplier,
the signal and image appear in Fig. 8b. Note that as the
signal is shifted to lower intensity values, the image
becomes darker (upper frame in Fig. 8b). When the gain
is adjusted to the full intensity range (Fig. 8c), the image
exhibits a significant amount of detail with good contrast
and high resolution.
Figure 8. Gain and offset control in confocal microscopy

photomultiplier detection units. The specimen is a living
adherent culture of Indian Muntjac deer skin fibroblast
cells treated with MitoTracker Red CMXRos. (a) The raw
confocal image (upper frame) along with the signal from
the photomultiplier. (b) Signal and confocal image after
applying a negative offset voltage to the photomultiplier.
(c) Final signal and image after the gain has been
adjusted to fill the entire intensity range.
The photomultiplier gain adjustment is utilized to electronically stretch the input signal by multiplying with a
constant factor prior to digitization by the analog-to-digital
converter (40). The result is a more complete representation of gray level values between black and white, and an
increase in apparent dynamic range. If the gain setting is
increased beyond the optimal point, the image becomes
grainy, but this maneuver is sometimes necessary to capture the maximum number of gray levels present in the
image. Advanced confocal microscopy software packages
ease the burden of gain and offset adjustment by using a
pseudocolor display function to associate pixel values with
gray levels on the monitor. For example, the saturated
pixels (255) can be displayed in yellow or red, while blackevel pixels (0) are shown in blue or green, with intermediate gray levels displayed in shades of gray representing
their true values. When the photomultiplier output is
properly adjusted, just a few red (or yellow) and blue (or
green) pixels are present in the image, indicating that the
full dynamic range of the photomultiplier is being utilized.
Established techniques in the field of enhanced night
vision have been applied with dramatic success to photomultipliers designed for confocal microscopy (63,64). Several manufacturers have collaborated to fabricate a headon photomultiplier containing a specialized prism system
that assists in the collection of photons. The prism operates
by diverting the incoming photons to a pathway that
promotes total internal reflection in the photomultiplier
envelope adjacent to the photocathode. This configuration
increases the number of potential interactions between the
photons and the photocathode, resulting in an increase in
quantum efficiency by more than a factor of 2 in the green
spectral region, 4 in the red region, and even higher in the
IR (59). Increasing the ratio of photoelectrons generated to
the number of incoming photons serves to increase the
electrical current from the photomultiplier, and to produce
a higher sensitivity for the instrument.
Photomultipliers are the ideal photometric detectors for
confocal microscopy due to their speed, sensitivity, high
signal/noise ratio, and adequate dynamic range (5961).
High end confocal microscope systems have several photomultipliers that enable simultaneous imaging of different
fluorophores in multiply labeled specimens. Often, an additional photomultiplier is included for imaging the specimen
with transmitted light using differential interference or
phase-contrast techniques. In general, confocal microscopes contain three photomultipliers for the fluorescence
color channels (red, green, and blue; each with a separate
pinhole aperture) utilized to discriminate between fluorophores, along with a fourth for transmitted or reflected
light imaging. Signals from each channel can be collected
simultaneously and the images merged into a single profile
that represents the real colors of the stained specimen. If
the specimen is also imaged with a brightfield contrastenhancing technique, such as differential interference contrast (65), the fluorophore distribution in the fluorescence
image can be overlaid onto the brightfield image to determine the spatial location of fluorescence emission within
the structural domains.
ACOUSTOOPTIC TUNABLE FILTERS IN CONFOCAL

MICROSCOPY
The integration of optoelectronic technology into confocal
microscopy has provided a significant enhancement in the
versatility of spectral control for a wide variety of fluorescence investigations. The acoustooptic tunable filter
(AOTF) is an electrooptical device that functions as an
electronically tunable excitation filter to simultaneously
modulate the intensity and wavelength of multiple laser
lines from one or more sources (66). Devices of this type rely
on a specialized birefringent crystal whose optical properties vary upon interaction with an acoustic wave. Changes
in the acoustic frequency alter the diffraction properties of
the crystal, enabling very rapid wavelength tuning, limited
only by the acoustic transit time across the crystal.
An acoustooptic tunable filter designed for microscopy
typically consists of a tellurium dioxide or quartz anisotropic crystal to which a piezoelectric transducer is bonded
(6770). In response to the application of an oscillating
radio frequency (RF) electrical signal, the transducer generates a high frequency vibrational (acoustic) wave that
propagates into the crystal. The alternating ultrasonic
acoustic wave induces a periodic redistribution of the
refractive index through the crystal that acts as a transmission diffraction grating or Bragg diffracter to deviate a
portion of incident laser light into a first-order beam, which
is utilized in the microscope (or two first-order beams when
the incident light is nonpolarized). Changing the frequency
of the transducer signal applied to the crystal alters the
period of the refractive index variation, and therefore, the
wavelength of light that is diffracted. The relative intensity
of the diffracted beam is determined by the amplitude
(power) of the signal applied to the crystal.
In the traditional fluorescence microscope configuration, including many confocal systems, spectral filtering
of both excitation and emission light is accomplished utilizing thin-film interference filters (7). These filters are
limiting in several respects. Because each filter has a fixed
central wavelength and passband, several filters must be
457
utilized to provide monochromatic illumination for multispectral imaging, as well as to attenuate the beam for
intensity control, and the filters are often mechanically
interchanged by a rotating turret mechanism. Interference
filter turrets and wheels have the disadvantages of limited
wavelength selection, vibration, relatively slow switching
speed, and potential image shift (70). They are also susceptible to damage and deterioration caused by exposure to
heat, humidity, and intense illumination, which changes
their spectral characteristics over time. In addition, the
utilization of filter wheels for illumination wavelength
selection has become progressively more complex and
expensive as the number of lasers being employed has
increased with current applications.
Rotation of filter wheels and optical block turrets introduces mechanical vibrations into the imaging and illumination system, which consequently requires a time delay
for damping of perhaps 50 ms, even if the filter transition
itself can be accomplished more quickly. Typical filter
change times are considerably slower in practice, however,
and range on the order of 0.10.5 s. Mechanical imprecision
in the rotating mechanism can introduce registration
errors when sequentially acquired multicolor images are
processed. Furthermore, the fixed spectral characteristics
of interference filters do not allow optimization for different
fluorophore combinations, nor for adaptation to new fluorescent dyes, limiting the versatility of both the excitation
and detection functions of the microscope. Introduction of
the AOTF to confocal systems overcomes most of the filter
wheel disadvantages by enabling rapid simultaneous electronic tuning and intensity control of multiple laser lines
from several lasers.
As applied in laser scanning confocal microscopy, one of
the most significant benefits of the AOTF is its capability to
replace much more complex and unwieldy filter mechanisms for controlling light transmission, and to apply intensity modulation for wavelength discrimination purposes
(67,70). The ability to perform extremely rapid adjustments in the intensity and wavelength of the diffracted
beam gives the AOTF unique control capabilities. By varying the illumination intensity at different wavelengths, the
response of multiple fluorophores, for example, can be
balanced for optimum detection and recording (71). In
addition, digital signal processors along with phase and
frequency lock-in techniques can be employed to discriminate emission from multiple fluorophores or to extract low
level signals from background.
A practical light source configuration scheme utilizing
an acoustooptic tunable filter for confocal microscopy is
illustrated in Fig. 9. The output of three laser systems
(violet diode, argon, and argonkrypton) are combined by
dichromatic mirrors and directed through the AOTF,
where the first-order diffracted beam (green) is collinear
and is launched into a single-mode fiber. The undiffracted
laser beams (violet, green, yellow, and red) exit the AOTF
at varying angles and are absorbed by a beam stop (not
illustrated). The major lines (wavelengths) produced by
each laser are indicated (in nm) beneath the hot and cold
mirrors. The dichromatic mirror reflects wavelengths
< 525 nm and transmits longer wavelengths. Two longer
wavelength lines produced by the argonkrypton laser
458
Figure 9. Configuration scheme utilizing an AOTF for laser

intensity control and wavelength selection in confocal microscopy.
(568 and 648 nm) are reflected by the hot mirror, while the
output of the argon laser (458, 476, 488, and 514 nm) is
reflected by the dichromatic mirror and combined with the
transmitted light from the argonkrypton laser. Output
from the violet diode laser (405 nm) is reflected by the cold
mirror and combined with the longer wavelengths from
the other two lasers, which are transmitted through the
mirror.
Because of the rapid optical response from the AOTF
crystal to the acoustic transducer, the acoustooptic interaction is subject to abrupt transitions resembling a rectangular rather than sinusoidal waveform (66). This
results in the occurrence of sidelobes in the AOTF
passband on either side of the central transmission peak.
Under ideal acoustooptic conditions, these sidelobes should
be symmetrical about the central peak, with the first lobe
having 4.7% of the central peaks intensity. In practice, the
sidelobes are commonly asymmetrical and exhibit other
deviations from predicted structure caused by variations in
the acoustooptic interaction, among other factors. In order
to reduce the sidelobes in the passband to insignificant
levels, several types of amplitude apodization of the acoustic
wave are employed (66,67), including various window functions, which have been found to suppress the highest sidelobe by 3040 dB. One method that can be used in reduction
of sidelobe level with noncollinear AOTFs is to apply spatial
apodization by means of weighted excitation of the transducer. In the collinear AOTF, a different approach has been
employed, which introduces an acoustic pulse, apodized in
time, into the filter crystal.
The effective linear aperture of an AOTF is limited by
the acoustic beam height in one dimension (ID) and by the
acoustic attenuation across the optical aperture (the acoustic transit distance) in the other dimension (67). The height
of the acoustic beam generated within the AOTF crystal is
determined by the performance and physical properties of
the acoustic transducer. Acoustic attenuation in crystalline
materials, such as tellurium dioxide, is proportional to
the square of acoustic frequency, and is therefore a more
problematic limitation to linear aperture size in the shorter
wavelength visible light range, which requires higher RF
frequencies for tuning. Near-IR and IR radiation produces
less restrictive limitations because of the lower acoustic

frequencies associated with diffraction of these longer
wavelengths.
The maximum size of an individual acoustic transducer
is constrained by performance and power requirements in
addition to the geometric limitations of the instrument
configuration, and AOTF designers may use an array of
transducers bonded to the crystal in order to increase the
effective lateral dimensions of the propagating acoustic
beam, and to enlarge the area of acoustooptic interaction
(66,67,70). The required drive power is one of the most
important variables in acoustooptic design, and generally
increases with optical aperture and for longer wavelengths. In contrast to acoustic attenuation, which is
reduced in the IR spectral range, the higher power required
to drive transducers for infrared AOTFs is one of the
greatest limitations in these devices. High drive power
levels result in heating of the crystal, which can cause
thermal drift and instability in the filter performance (66).
This is particularly a problem when acoustic power and
frequency are being varied rapidly over a large range, and
the crystal temperature does not have time to stabilize,
producing transient variations in refractive index. If an
application requires wavelength and intensity stability
and repeatability, the AOTF should be maintained at a
constant temperature. One approach taken by equipment
manufacturers to minimize this problem is to heat the
crystal above ambient temperature, to a level at which it
is relatively unaffected by the additional thermal input of
the transducer drive power. An alternative solution is to
house the AOTF in a thermoelectrically cooled housing
that provides precise temperature regulation. Continuing
developmental efforts promise to lead to new materials
that can provide relatively large apertures combined with
effective separation of the filtered and unfiltered beams
without use of polarizers, while requiring a fraction of the
typical device drive power.
In a noncollinear AOTF, which spatially separates the
incident and diffracted light paths, the deflection angle (the
angle separating diffracted and undiffracted light beams
exiting the crystal) is an additional factor limiting the
effective aperture of the device (67). As discussed previously, the deflection angle is greater for crystals having
greater birefringence, and determines in part the propagation distance required for adequate separation of the diffracted and undiffracted beams to occur after exiting the
crystal. The required distance is increased for larger
entrance apertures, and this imposes a practical limit on
maximum aperture size because of constraints on the
physical dimensions of components that can be incorporated into a microscope system. The angular aperture is
related to the total light collecting power of the AOTF, an
important factor in imaging systems, although in order to
realize the full angular aperture without the use of polarizers in the noncollinear AOTF, its value must be smaller
than the deflection angle. Because the acoustooptic tunable
filter is not an image-forming component of the microscope
system (it is typically employed for source filtering), there
is no specific means of evaluating the spatial resolution
for this type of device (70). However, the AOTF may restrict
the attainable spatial resolution of the imaging system
because of its limited linear aperture size and acceptance

angle, in the same manner as other optical components.
Based on the Rayleigh criterion and the angular and linear
apertures of the AOTF, the maximum number of resolvable
image elements may be calculated for a given wavelength,
utilizing different expressions for the polar and azimuthal
planes. Although diffraction limited resolution can be
attained in the azimuthal plane, dispersion in the AOTF
limits the resolution in the polar plane, and measures must
be taken to suppress this factor for optimum performance.
The dependence of deflection angle on wavelength can
produce one form of dispersion, which is typically negligible when tuning is performed within a relatively narrow
bandwidth, but significant in applications involving operation over a broad spectral range. Changes in deflection
angle with wavelength can result in image shifts during
tuning, producing errors in techniques, such as ratio imaging of fluorophores excited at different wavelengths, and
in other multispectral applications. When the image shift
obeys a known relationship to wavelength, corrections can
be applied through digital processing techniques (1,7).
Other effects of dispersion, including reduced angular
resolution, may result in image degradation, such as blurring, that requires more elaborate measures to suppress.
SUMMARY OF AOTF BENEFITS IN CONFOCAL

MICROSCOPY
Considering the underlying principles of operation and
performance factors that relate to the application of AOTFs
in imaging systems, a number of virtues from such devices
for light control in fluorescence confocal microscopy are
apparent. Several benefits of the AOTF combine to greatly
459
enhance the versatility of the latest generation of confocal

instruments, and these devices are becoming increasing
popular for control of excitation wavelength ranges and
intensity. The primary characteristic that facilitates
nearly every advantage of the AOTF is its capability to
allow the microscopist control of the intensity and/or illumination wavelength on a pixel-by-pixel basis while maintaining a high scan rate (7). This single feature translates
into a wide variety of useful analytical microscopy tools,
which are even further enhanced in flexibility when laser
illumination is employed.
One of the most useful AOTF functions allows the
selection of small user-defined specimen areas (commonly
termed regions of interest; ROI) that can be illuminated
with either greater or lesser intensity, and at different
wavelengths, for precise control in photobleaching techniques, excitation ratio studies, resonance energy-transfer
investigations, or spectroscopic measurements (see Fig. 10).
The illumination intensity can not only be increased in
selected regions for controlled photobleaching experiments
(7173), but can be attenuated in desired areas in order to
minimize unnecessary photobleaching. When the illumination area is under AOTF control, the laser exposure is
restricted to the scanned area by default, and the extremely
rapid response of the device can be utilized to provide beam
blanking during the flyback interval of the galvanometer
scanning mirror cycle, further limiting unnecessary specimen exposure. In practice, the regions of excitation are
typically defined by freehand drawing or using tools to
produce defined geometrical shapes in an overlay plane
on the computer monitor image. Some systems allow any
number of specimen areas to be defined for laser exposure,
and the laser intensity to be set to different levels for each
area, in intensity increments as small as 0.1%. When the
Figure 10. AOTF selection of specific regions for excitation in

confocal microscopy. (a) Region of Interest (ROI) selected for
fluorescence recovery after photobleaching (FRAP) experiments. (b) Freehand ROIs for selective excitation. (c) ROI for
fluorescence resonance energy-transfer (FRET) analysis. (d) ROI
for photoactivation and photoconversion of fluorescent proteins.
460
AOTF is combined with multiple lasers and software that

allows time course control of sequential observations, timelapse experiments can be designed to acquire data from
several different areas in a single experiment, which might,
for example, be defined to correspond to different cellular
organelles.
Figure 10 illustrates several examples of several userdefined ROIs that were created for advanced fluorescence
applications in laser scanning confocal microscopy. In each
image, the ROI is outlined with a yellow border. The rat
kangaroo kidney epithelial cell (PtK2 line) presented in
Fig. 10a has a rectangular area in the central portion of the
cytoplasm that has been designated for photobleaching
experiments. Fluorophores residing in this region can be
selectively destroyed by high power laser intensity, and the
subsequent recovery of fluorescence back into the photobleached region monitored for determination of diffusion
coefficients. Several freehand ROIs are illustrated in
Fig. 10b, which can be targets for selective variation of
illumination intensities or photobleaching and photoactivation experiments. Fluorescence resonance energytransfer emission ratios can be readily determined using
selected regions in confocal microscopy by observing the
effect of bleaching the acceptor fluorescence in these areas
(Fig. 10c; African green monkey kidney epithelial cells
labeled with Cy3 and Cy5 conjugated to cholera toxin,
which localizes in the plasma membrane). The AOTF control of laser excitation in selected regions with confocal
microscopy is also useful for investigations of protein diffusion in photoactivation studies (7476) using fluorescent
proteins, as illustrated in Fig. 10d. This image frame
presents the fluorescence emission peak of the Kaede
protein as it shifts from green to red in HeLa (human
Figure 11. Fluorophore bleedthrough control with neutral

density filters and sequential scanning using AOTF laser
modulation. Adherent human lung fibroblast (MRC-5 line)
cells were stained with Texas Red conjugated to phalloidin
(actin; red) and counterstained with SYTOX green (nuclei;
green). (a) Typical cell imaged with neutral density filters. (b)
The same cell imaged using sequential line scanning controlled
by an AOTF laser combiner. (c) and (d) Colocalization scatterplots
derived from the images in (a) and (b), respectively.
cervical carcinoma) cell nuclei using selected illumination

(yellow box) with a 405 nanometer violetblue diode laser.
The rapid intensity and wavelength switching capabilities of the AOTF enable sequential line scanning of multiple laser lines to be performed in which each excitation
wavelength can be assigned a different intensity in order to
balance the various signal levels for optimum imaging (77).
Sequential scanning of individual lines minimizes the time
differential between signal acquisitions from the various
fluorophores while reducing crossover, which can be a significant problem with simultaneous multiple-wavelength
excitation (Fig. 11). The synchronized incorporation of
multiple fluorescent probes into living cells has grown
into an extremely valuable technique for study of protein
protein interactions, and the dynamics of macromolecular
complex assembly. The refinement of techniques for
incorporating green fluorescent protein (GFP) and its
numerous derivatives into the protein-synthesizing
mechanisms of the cell has revolutionized living cell
experimentation (7880). A major challenge in multipleprobe studies using living tissue is the necessity to acquire
the complete multispectral data set quickly enough to
minimize specimen movement and molecular changes
that might distort the true specimen geometry or dynamic
sequence of events (3234). The AOTF provides the speed
and versatility to control the wavelength and intensity
illuminating multiple specimen regions, and to simultaneously or sequentially scan each at sufficient speed to
accurately monitor dynamic cellular processes.
A comparison between the application of AOTFs and
neutral density filters (78) to control spectral separation of
fluorophore emission spectra in confocal microscopy is
presented in Fig. 11. The specimen is a monolayer culture
of adherent human lung fibroblast (MRC-5 line) cells

stained with Texas Red conjugated to phalloidin (targeting
the filamentous actin network) and SYTOX Green (staining DNA in the nucleus). A neutral density filter that
produces the high excitation signals necessary for both
fluorophores leads to a significant amount of bleedthrough
of the SYTOX Green emission into the Texas Red channel
(Fig. 11a; note the yellow nuclei). The high degree of
apparent colocalization between SYTOX Green and Texas
Red is clearly illustrated by the scatterplot in Fig. 11b. The
two axes in the scatterplot represent the SYTOX Green
(abscissa) and the Texas Red (ordinate) channels. In order
to balance the excitation power levels necessary to selectively illuminate each fluorophore with greater control of
emission intensity, an AOTF was utilized to selectively
reduce the SYTOX Green excitation power (Argon-ion laser
line at 488 nm). Note the subsequent reduction in bleedthrough as manifested by green color in the cellular nuclei
in Fig. 11c. The corresponding scatterplot (Fig. 11d) indicates a dramatically reduced level of bleed-through (and
apparent colocalization) of SYTOX Green into the Texas
Red channel.
The development of the AOTF has provided substantial
additional versatility to techniques, such as fluorescence
recovery after photobleaching (FRAP; 81,82), fluorescence
loss in photobleaching (FLIP; 83), as well as in localized
photoactivated fluorescence (uncaging; 84) studies
(Fig. 10). The FRAP technique (81,82) was originally conceived to measure diffusion rates of fluorescently tagged
proteins in organelles and cell membranes. In the conventional FRAP procedure, a small spot on the specimen is
continuously illuminated at a low light flux level and the
emitted fluorescence is measured. The illumination level is
then increased to a very high level for a brief time to
destroy the fluorescent molecules in the illuminated region
by rapid bleaching. After the light intensity is returned to
the original low level, the fluorescence is monitored to
determine the rate at which new unbleached fluorescent
molecules diffuse into the depleted region. The technique,
as typically employed, has been limited by the fixed geometry of the bleached region, which is often a diffractionlimited spot, and by having to mechanically adjust the
illumination intensity (using shutters or galvanometerdriven components). The AOTF not only allows nearinstantaneous switching of light intensity, but also can be
utilized to selectively bleach randomly specified regions of
irregular shape, lines, or specific cellular organelles, and to
determine the dynamics of molecular transfer into the region.
By enabling precise control of illuminating beam geometry and rapid switching of wavelength and intensity,
the AOTF is a significant enhancement to application of
the FLIP technique in measuring the diffusional mobility
of certain cellular proteins (83). This technique monitors
the loss of fluorescence from continuously illuminated
localized regions and the redistribution of fluorophore
from distant locations into the sites of depletion. The data
obtained can aid in the determination of the dynamic
interrelationships between intracellular and intercellular
components in living tissue, and such fluorescence loss
studies are greatly facilitated by the capabilities of the
AOTF in controlling the microscope illumination.
461
The method of utilizing photoactivated fluorescence has

been very useful in studies, such as those examining the
role of calcium ion concentration in cellular processes, but
has been limited in its sensitivity to localized regional
effects in small organelles or in close proximity to cell
membranes. Typically, fluorescent species that are inactivated by being bound to a photosensitive species (referred
to as being caged) are activated by intense illumination
that frees them from the caging compound and allows them
to be tracked by the sudden appearance of fluorescence
(84). The use of the AOTF has facilitated the refinement of
such studies to assess highly localized processes such as
calcium ion mobilization near membranes, made possible
because of the precise and rapid control of the illumination
triggering the activation (uncaging) of the fluorescent
molecule of interest.
Because the AOTF functions, without use of moving
mechanical components, to electronically control the wavelength and intensity of multiple lasers, great versatility is
provided for external control and synchronization of laser
illumination with other aspects of microscopy experiments.
When the confocal instrument is equipped with a controller
module having input and output trigger terminals, laser
intensity levels can be continuously monitored and
recorded, and the operation of all laser functions can be
controlled to coordinate with other experimental specimen
measurements, automated microscope stage movements,
sequential time-lapse recording, and any number of other
operations.
RESOLUTION AND CONTRAST IN CONFOCAL

MICROSCOPY
All optical microscopes, including conventional widefield,
confocal, and two-photon instruments are limited in the
resolution that they can achieve by a series of fundamental
physical factors (1,3,57,24,8589). In a perfect optical
system, resolution is restricted by the numerical aperture
of optical components and by the wavelength of light, both
incident (excitation) and detected (emission). The concept
of resolution is inseparable from contrast, and is defined as
the minimum separation between two points that results in
a certain level of contrast between them (24). In a typical
fluorescence microscope, contrast is determined by the number of photons collected from the specimen, the dynamic
range of the signal, optical aberrations of the imaging
system, and the number of picture elements (pixels) per
unit area in the final image (66,8688).
The influence of noise on the image of two closely spaced
small objects is further interconnected with the related
factors mentioned above, and can readily affect the quality
of resulting images (29). While the effects of many instrumental and experimental variables on image contrast, and
consequently on resolution, are familiar and rather obvious,
the limitation on effective resolution resulting from the
division of the image into a finite number of picture elements
(pixels) may be unfamiliar to those new to digital microscopy. Because all digital confocal images employing laser
scanners and/or camera systems are recorded and processed
in terms of measurements made within discrete pixels (66),
462
some discussion of the concepts of sampling theory is

required. This is appropriate to the subject of contrast
and resolution because it has a direct bearing on the ability
to record two closely spaced objects as being distinct.
In addition to the straightforward theoretical aspects of
resolution, regardless of how it is defined, the reciprocal
relationship between contrast and resolution has practical
significance because the matter of interest to most microscopists is not resolution, but visibility. The ability to
recognize two closely spaced features as being separate
relies on advanced functions of the human visual system to
interpret intensity patterns, and is a much more subjective
concept than the calculation of resolution values based on
diffraction theory (24). Experimental limitations and the
properties of the specimen itself, which vary widely, dictate
that imaging cannot be performed at the theoretical maximum resolution of the microscope.
The relationship between contrast and resolution with
regard to the ability to distinguish two closely spaced
specimen features implies that resolution cannot be
defined without reference to contrast, and it is this interdependence that has led to considerable ambiguity involving the term resolution and the factors that influence it in
microscopy (29). As discussed above, recent advances in
fluorescent protein technology have led to an enormous
increase in studies of dynamic processes in living cells and
tissues (7176,7883). Such specimens are optically thick
and inhomogeneous, resulting in a far-from-ideal imaging
situation in the microscope. Other factors, such as cell
viability and sensitivity to thermal damage and photobleaching, place limits on the light intensity and duration
of exposure, consequently limiting the attainable resolution. Given that the available timescale may be dictated by
these factors and by the necessity to record rapid dynamic
events in living cells, it must be accepted that the quality of
images will not be as high as those obtained from fixed and
stained specimens. The most reasonable resolution goal for
imaging in a given experimental situation is that the
microscope provides the best resolution possible within
the constraints imposed by the experiment.
Figure 12. Schematic diagram of an Airy disk diffraction

pattern and the corresponding three-dimensional point spread
functions for image formation in confocal microscopy. Intensity
profiles of a single Airy disk, as well as the first and higher order
maxima are illustrated in the graphs.
THE AIRY DISK AND LATERAL RESOLUTION

Imaging a point-like light source in the microscope produces an electromagnetic field in the image plane whose
amplitude fluctuations can be regarded as a manifestation
of the response of the optical system to the specimen. This
field is commonly represented through the amplitude
point spread function, and allows evaluation of the
optical transfer properties of the combined system components (29,8688). Although variations in field amplitude
are not directly observable, the visible image of the point
source formed in the microscope and recorded by its
imaging system is the intensity point spread function,
which describes the system response in real space. Actual
specimens are not point sources, but can be regarded as a
superposition of an infinite number of objects having
dimensions below the resolution of the system. The properties of the intensity point spread function (PSF; see Fig. 12)
in the image plane as well as in the axial direction are
major factors in determining the resolution of a microscope
(1,24,29,40,8589).
It is possible to experimentally measure the intensity
point spread function in the microscope by recording the
image of a subresolution spherical bead as it is scanned
through focus (a number of examples may be found in the
literature). Because of the technical difficulty posed in
direct measurement of the intensity point spread function,
calculated point spread functions are commonly utilized to
evaluate the resolution performance of different optical
systems, as well as the optical-sectioning capabilities of
confocal, two-photon, and conventional widefield microscopes. Although the intensity point spread function
extends in all three dimensions, with regard to the relationship between resolution and contrast, it is useful to
consider only the lateral components of the intensity distribution, with reference to the familiar Airy disk (24).
The intensity distribution of the point-spread function
in the plane of focus is described by the rotationally symmetric Airy pattern. Because of the cylindrical symmetry
of the microscope lenses, the two lateral components
(x and y) of the Airy pattern are equivalent, and the

pattern represents the lateral intensity distribution as
a function of distance from the optical axis (24). The
lateral distance is normalized by the numerical aperture
of the system and the wavelength of light, and therefore is
dimensionless. Figure 12 (Airy disk and intensity function) illustrates diagrammatically the formation and
characteristics of the Airy disk, the related 3D point
spread function, and Airy patterns in the fluorescence
microscope. Following the excitation of fluorophores in a
point-like specimen region, fluorescence emission occurs
in all directions, a small fraction of which is selected and
focused by the optical components into an image plane
where it forms an Airy disk surrounded by concentric
rings of successively decreasing maximum and minimum
intensity (the Airy pattern). The Airy pattern intensity
distribution is the result of Fraunhofer diffraction of light
passing through a circular aperture, and in a perfect
optical system exhibits a central intensity maximum
and higher order maxima separated by regions of zero
intensity (85). The distance of the zero crossings from the
optical axis, when the distance is normalized by the
numerical aperture and wavelength, occur periodically
(Fig. 12). When the intensity on the optical axis is normalized to one (100%), the proportional heights of the first
four higher order maxima are 1.7%, 0.4%, 0.2%, and
0.08%, respectively.
A useful approach to the concept of resolution is based
on consideration of an image formed by two point-like
objects (specimen features), under the assumption that
the image-forming process is incoherent, and that the
interaction of the separate object images can be described
using intensity point spread functions. The resulting image
is then composed of the sum of two Airy disks, the characteristics of which depend on the separation distance
between the two points (24,86). When sufficiently separated, the intensity change in the area between the objects
is the maximum possible, cycling from the peak intensity
(at the first point) to zero and returning to the maximum
value at the center of the second point. At decreased
distance in object space, the intensity distribution functions of the two points, in the image plane, begin to overlap
and the resulting image may appear to be that of a single
larger or brighter object or feature rather than being
recognizable as two objects. If resolution is defined, in
general terms, as the minimum separation distance at
which the two objects can be sufficiently distinguished,
it is obvious that this property is related to the width of the
intensity peaks (the point spread function). Microscope
resolution is directly related, therefore, to the full-width
at half maximum (fwhm) of the instruments intensity point
spread function in the component directions (29,86,87).
Some ambiguity in use of the term resolution results
from the variability in defining the degree of separation
between features and their point spread functions that is
sufficient to allow them to be distinguished as two objects
rather than one. In general, minute features of interest in
microscopy specimens produce point images that overlap to
some extent, displaying two peaks separated by a gap
(1,24,29,40,86). The greater the depth of the gap between
the peaks, the easier it is to distinguish, or resolve, the two
463
objects. By specifying the depth of the dip in intensity

between two overlapping point spread functions, the ambiguity in evaluating resolution can be removed, and a
quantitative aspect introduced.
In order to quantify resolution, the concept of contrast
is employed, which is defined for two objects of equal
intensity as the difference between their maximum intensity and the minimum intensity occurring in the space
between them (55,86,89). Because the maximum intensity
of the Airy disk is normalized to one, the highest achievable
contrast is also one, and occurs only when the spacing
between the two objects is relatively large, with sufficient
separation to allow the first zero crossing to occur in their
combined intensity distribution. At decreased distance, as
the two point spread functions begin to overlap, the dip in
intensity between the two maxima (and the contrast) is
increasingly reduced. The distance at which two peak
maxima are no longer discernible, and the contrast
becomes zero, is referred to as the contrast cut-off distance
(24,40). The variation of contrast with distance allows
resolution, in terms of the separation of two points, to be
defined as a function of contrast.
The relationship between contrast and separation distance for two point-like objects is referred to as the contrast/distance function or contrast transfer function
(31,90). Resolution can be defined as the separation distance at which two objects are imaged with a certain
contrast value. It is obvious that when zero contrast exists,
the points are not resolved; the so-called Sparrow criterion defines the resolution of an optical system as being
equivalent to the contrast cut-off distance (24). It is common, however, to specify that greater contrast is necessary
to adequately distinguish two closely spaced points
visually, and the well-known Rayleigh criterion (24) for
resolution states that two points are resolved when the
first minimum (zero crossing) of one Airy disk is aligned
with the central maximum of the second Airy disk. Under
optimum imaging conditions, the Rayleigh criterion
separation distance corresponds to a contrast value of
26.4%. Although any contrast value >0 can be specified
in defining resolution, the 26% contrast of the Rayleigh
criterion is considered reasonable in typical fluorescence
microscopy applications, and is the basis for the common
expression defining lateral resolution according to the
following equation (24), in which the point separation (r)
in the image plane is the distance between the central
maximum and the first minimum in the Airy disk:
rlateral 1:22 l=2NA 0:6 l=NA
where l is the emitted light wavelength and NA is the
numerical aperture of the objective.
Resolution in the microscope is directly related to the
fwhm dimensions of the microscopes point spread function,
and it is common to measure this value experimentally in
order to avoid the difficulty in attempting to identify intensity maxima in the Airy disk. Measurements of resolution
utilizing the fwhm values of the point spread function are
somewhat smaller than those calculated employing the
Rayleigh criterion. Furthermore, in confocal fluorescence
configurations, single-point illumination scanning and
464
single-point detection are employed, so that only the

fluorophores in the shared volume of the illumination
and detection point spread functions are able to be
detected. The intensity point spread function in the confocal case is, therefore, the product of the independent
illumination intensity and detection intensity point
spread functions. For confocal fluorescence, the lateral
(and axial) extent of the point spread function is reduced
by 30% compared to that in the wide-field microscope.
Because of the narrower intensity point spread function,
the separation of points required to produce acceptable
contrast in the confocal microscope (29,31) is reduced to a
distance approximated by
most applicable to fluorescence emission are similar in

form to the expressions evaluating depth of field, and
demonstrate that axial resolution is proportional to the
wavelength, and refractive index of the specimen medium,
and inversely proportional to the square of the numerical
aperture. Consequently, the NA of the microscope objective
has a much greater effect on axial resolution than does the
emission wavelength. One equation (89) commonly used to
describe axial resolution for the confocal configuration is
given below, with h representing the index of refraction,
and the other variables as specified previously:
rlateral 0:4l=NA
Although the confocal microscope configuration exhibits

only a modest improvement in measured axial resolution
over that of the widefield microscope, the true advantage of
the confocal approach is in the optical sectioning capability
in thick specimens, which results in a dramatic improvement in effective axial resolution over conventional techniques. The optical sectioning properties of the confocal
microscope result from the characteristics of the integrated
intensity point spread function, which has a maximum in
the focal plane when evaluated as a function of depth. The
equivalent integral of intensity point spread function for the
conventional widefield microscope is constant as a function
of depth, producing no optical sectioning capabilities.
If the illumination and fluorescence emission wavelengths are approximately the same, the confocal fluorescence microscope Airy disk size is the square of the
wide-field microscope Airy disk. Consequently, the contrast cut-off distance is reduced in the confocal arrangement, and equivalent contrast can be achieved at a shorter
distance compared to the widefield illumination configuration. Regardless of the instrument configuration, the
lateral resolution displays a proportional relationship to
wavelength, and is inversely proportional to the objective
lens numerical aperture.
As noted previously, lateral resolution is of primary
interest in discussing resolution and contrast, although
the axial extent of the microscope intensity point spread
function is similarly reduced in the confocal arrangement
as compared to the widefield fluorescence configuration
(86,89). Reasonable contrast between point-like objects
lying on the optical axis occurs when they are separated
by the distance between the central maximum and the first
minimum of the axial point spread function component.
Figure 13 presents the axial intensity distributions (89)
for a typical widefield (Fig. 13a) and confocal (Fig. 13b)
fluorescence microscope. Note the dramatic reduction in
intensity of the wings in the confocal distribution as a
function of distance from the central maximum.
A variety of equations are presented in the literature
that pertains to different models for calculating axial
resolution for various microscope configurations. The ones
Figure 13. Comparison of axial (xz) point spread functions for

widefield (left) and confocal (right) microscopy.
raxial 1:4 lh=NA2
FLUOROPHORES FOR CONFOCAL MICROSCOPY

Biological laser scanning confocal microscopy relies heavily
on fluorescence as an imaging mode, primarily due to the
high degree of sensitivity afforded by the technique coupled
with the ability to specifically target structural components
and dynamic processes in chemically fixed as well as living
cells and tissues. Many fluorescent probes are constructed
around synthetic aromatic organic chemicals designed
to bind with a biological macromolecule (e.g., a protein
or nucleic acid) or to localize within a specific structural
region, such as the cytoskeleton, mitochondria, Golgi apparatus, endoplasmic reticulum, and nucleus (90). Other
probes are employed to monitor dynamic processes and
localized environmental variables, including concentrations of inorganic metallic ions, pH, reactive oxygen species, and membrane potential (91). Fluorescent dyes are
also useful in monitoring cellular integrity (live versus
dead and apoptosis), endocytosis, exocytosis, membrane
fluidity, protein trafficking, signal transduction, and enzymatic activity (92). In addition, fluorescent probes have
been widely applied to genetic mapping and chromosome
analysis in the field of molecular genetics.
The history of synthetic fluorescent probes dates back
over a century to the late-1800s when many of the cornerstone dyes for modern histology were developed. Among
these were pararosaniline, methyl violet, malachite green,
safranin O, methylene blue, and numerous azo (nitrogen)
dyes, such as Bismarck brown (93). Although these dyes
were highly colored and capable of absorbing selected
bands of visible light, most were only weakly fluorescent
and would not be useful for the fluorescence microscopes
that would be developed several decades later. However,
several synthetic dye classes synthesized during this period,

based on the xanthene and acridine heterocyclic ring systems, proved to be highly fluorescent and provided a foundation for the development of modern synthetic fluorescent
probes. Most notable among these early fluorescent dyes
were the substituted xanthenes, fluorescein and rhodamine
B, and the biaminated acridine derivative, acridine orange.
Fluorochromes were introduced to fluorescence microscopy in the early twentieth century as vital stains for
bacteria, protozoa, and trypanosomes, but did not see
widespread use until the 1920s when fluorescence microscopy was first used to study dye binding in fixed tissues
and living cells (7,93). However, it was not until the early
1940s that Coons developed a technique for labeling antibodies with fluorescent dyes, thus giving birth to the field of
immunofluorescence (94). Over the past 60 years, advances
in immunology and molecular biology have produced a wide
spectrum of secondary antibodies and provided insight into
the molecular design of fluorescent probes targeted at specific regions within macromolecular complexes.
Fluorescent probe technology and cell biology were
dramatically altered by the discovery of the GFP from
jellyfish and the development of mutant spectral variants,
which have opened the door to noninvasive fluorescence
multicolor investigations of subcellular protein localization, intermolecular interactions, and trafficking using
living cell cultures (79,80,95). More recently, the development of nanometer-sized fluorescent semiconductor quantum dots has provided a new avenue for research in
confocal and widefield fluorescence microscopy (96).
Despite the numerous advances made in fluorescent dye
synthesis during the past few decades, there is very little
solid evidence about molecular design rules for developing
new fluorochromes, particularly with regard to matching
absorption spectra to available confocal laser excitation
wavelengths. As a result, the number of fluorophores that
have found widespread use in confocal microscopy is a
limited subset of the many thousands that have been
discovered.
BASIC CHARACTERISTICS OF FLUOROPHORES

Fluorophores are catalogued and described according to
their absorption and fluorescence properties, including the
spectral profiles, wavelengths of maximum absorbance and
emission, and the fluorescence intensity of the emitted
light (92). One of the most useful quantitative parameters
for characterizing absorption spectra is the molar extinction coefficient (denoted with the Greek symbole, see
Fig. 14a), which is a direct measure of the ability of a
molecule to absorb light. The extinction coefficient is useful
for converting units of absorbance into units of molar
concentration, and is determined by measuring the absorbance at a reference wavelength (usually the maximum,
characteristic of the absorbing species) for a molar concentration in a defined optical path length. The quantum yield
of a fluorochrome or fluorophore represents a quantitative
measure of fluorescence emission efficiency, and is expressed
as the ratio of the number of photons emitted to the
number of photons absorbed. In other words, the quantum
465
Figure 14. Fluorescent spectral profiles, plotted as normalized

absorption or emission as a function of wavelength, for popular
synthetic fluorophores emitting in the blue, green, and red regions
of the visible spectrum. Each profile is identified with a colored
bullet in (a), which illustrates excitation spectra. (b) The emission
spectra for the fluorophores according to the legend in (a).
yield represents the probability that a given excited fluorochrome will produce an emitted (fluorescence) photon.
Quantum yields typically range between a value of 0 and 1
and fluorescent molecules commonly employed as probes
in microscopy have quantum yields ranging from very low
(0.05 or less) to almost unity. In general, a high quantum
yield is desirable in most imaging applications. The quantum yield of a given fluorophore varies, sometimes to large
extremes, with environmental factors, such as metallic
ion concentration, pH, and solvent polarity (92).
In most cases, the molar extinction coefficient for photon
absorption is quantitatively measured and expressed at a
specific wavelength, whereas the quantum efficiency is an
assessment of the total integrated photon emission over the
entire spectral band of the fluorophore (Fig. 14b). As
opposed to traditional arc-discharge lamps used with the
shortest range (1020 nm) bandpass interference filters in
wide-field fluorescence microscopy, the laser systems used
for fluorophore excitation in scanning confocal microscopy
restrict excitation to specific laser spectral lines that
encompass only a few nanometers (1,7). The fluorescence
emission spectrum for both techniques, however, is controlled by similar bandpass or longpass filters that can
cover tens to hundreds of nanometers (7). Below saturation
levels, fluorescence intensity is proportional to the product
of the molar extinction coefficient and the quantum yield of
the fluorophore, a relationship that can be utilized to judge
the effectiveness of emission as a function of excitation
wavelength(s). These parameters display an approximate
20-fold range in variation for the popular fluorophores
commonly employed for investigations in confocal microscopy with quantum yields ranging from 0.05 to 1.0, and
extinction coefficients ranging from 10,000 to 0.25 million
(L mol1). In general, the absorption spectrum of a fluorophore is far less dependent on environmental conditions
than the fluorescence emission characteristics (spectral
wavelength profile and quantum yield; 92).
Fluorophores chosen for confocal applications must
exhibit a brightness level and signal persistence sufficient
for the instrument to obtain image data that does not suffer
from excessive photobleaching artifacts and low signal/
noise ratios. In widefield fluorescence microscopy, excitation illumination levels are easily controlled with neutral
466
Table 1. Laser and Arc-Discharge Spectral Lines in Widefield and Confocal Microscopy
Laser Type
Argon-ion
Blue diode
Diode-pumped solid state
Heliumcadmium
Kryptonargon
Green heliumneon
Yellow heliumneon
Orange heliumneon
Red helium-neon
Red diode
Mercury arc
Xenon arc
Ultraviolet
Violet
351, 364
355
322, 354
405, 440
430, 442
442
Blue
Green
457, 477, 488
514
457, 473
532
488
Yellow
Orange
Red
561
568
647
543
594
612
633
635, 650
365
405, 436
467
density filters (40), and the intensity can be reduced

(coupled with longer emission signal collection periods)
to avoid saturation and curtail irreversible loss of fluorescence. Excitation conditions in confocal microscopy are
several orders of magnitude more severe, however, and
restrictions imposed by characteristics of the fluorophores
and efficiency of the microscope optical system become the
dominating factor in determining excitation rate and emission collection strategies (1,7,92).
Because of the narrow and wavelength-restricted laser
spectral lines employed to excite fluorophores in confocal
microscopy (Table 1), fluorescence emission intensity can
be seriously restricted due to poor overlap of the excitation
wavelengths with the fluorophore absorption band. In
addition, the confocal pinhole aperture, which is critical
in obtaining thin optical sections at high signal/noise
ratios, is responsible for a 2550% loss of emission intensity, regardless of how much effort has been expended on
fine-tuning and alignment of the microscope optical system
(7). Photomultiplier tubes are the most common detectors
in confocal microscopy, but suffer from a quantum efficiency that varies as a function of wavelength (especially in
the red and IR regions), further contributing to a wavelength-dependent loss of signal across the emission spectrum (5962). Collectively, the light losses in confocal
microscopy can result in a reduction of intensity exceeding
50 times of the level typically observed in widefield fluorescence instruments. It should be clear from the preceding
argument that fluorophore selection is one of the most
critical aspects of confocal microscopy, and instrumental
efficiency must be carefully considered, as well, in order to
produce high quality images.
In confocal microscopy, irradiation of the fluorophores
with a focused laser beam at high power densities increases
the emission intensity up to the point of dye saturation, a
condition whose parameters are dictated by the excited
state lifetime (97). In the excited state, fluorophores are
unable to absorb another incident photon until they emit a
lower energy photon through the fluorescence process.
When the rate of fluorophore excitation exceeds the rate
of emission decay, the molecules become saturated and the
ground state population decreases. As a result, a majority
of the laser energy passes through the specimen undiminished and does not contribute to fluorophore excitation.
Balancing fluorophore saturation with laser light intensity
546
579
levels is, therefore, a critical condition for achieving the optimal signal/noise ratio in confocal experiments (1,7,92,97). The
number of fluorescent probes currently available for confocal microscopy runs in the hundreds (90,93), with many
dyes having absorption maxima closely associated with
common laser spectral lines (90). An exact match between
a particular laser line and the absorption maximum of a
specific probe is not always possible, but the excitation
efficiency of lines near the maximum is usually sufficient
to produce a level of fluorescence emission that can be
readily detected.
Instrumentally, fluorescence emission collection can be
optimized by careful selection of objectives, detector aperture dimensions, dichromatic and barrier filters, as well as
maintaining the optical train in precise alignment (63). In
most cases, low magnification objectives with a high
numerical aperture should be chosen for the most demanding imaging conditions because light collection intensity
increases as the fourth power of the numerical aperture,
but only decreases as the square of the magnification.
However, the most important limitations in light collection
efficiency in confocal microscopy arise from restrictions
imposed by the physical properties of the fluorophores
themselves. As previously discussed, fluorescent probe
development is limited by a lack of knowledge of the specific
molecular properties responsible for producing optimum
fluorescence characteristics, and the design rules are insufficiently understood to be helpful as a guide to the development of more efficient fluorophores. The current success
in development of new fluorescent probes capable of satisfactory performance in confocal microscopy is a testament
to the progress made through use of empirical data and
assumptions about molecular structure extrapolated from
the properties of existing dyes, many of which were first
synthesized over a hundred years ago.
TRADITIONAL FLUORESCENT DYES

The choice of fluorescent probes for confocal microscopy
must address the specific capabilities of the instrument to
excite and detect fluorescence emission in the wavelength
regions made available by the laser systems and detectors.
Although the current lasers used in confocal microscopy
(Table 1) produce discrete lines in the UV, visible, and
near-IR portions of the spectrum, the location of these

spectral lines does not always coincide with absorption
maxima of popular fluorophores. In fact, it is not necessary
for the laser spectral line to correspond exactly with the
fluorophore wavelength of maximum absorption, but the
intensity of fluorescence emission is regulated by the fluorophore extinction coefficient at the excitation wavelength
(as discussed above). The most popular lasers for confocal
microscopy are air-cooled argon and kryptonargon ion
lasers, the new blue diode lasers, and a variety of helium
neon systems (7,40). Collectively, these lasers are capable of
providing excitation at 1012 specific wavelengths between
400 and 650 nm.
Many of the classical fluorescent probes that have been
successfully utilized for many years in widefield fluorescence (92,93), including fluorescein isothiocyanate, Lissamine rhodamine, and Texas red, are also useful in confocal
microscopy. Fluorescein is one of the most popular fluorochromes ever designed, and has enjoyed extensive application in immunofluorescence labeling. This xanthene dye
has an absorption maximum at 495 nm, which coincides
quite well with the 488 nm (blue) spectral line produced by
argon-ion and kryptonargon lasers, as well as the 436 and
467 principal lines of the mercury and xenon arc-discharge
lamps, respectively. In addition, the quantum yield of
fluorescein is very high and a significant amount of information has been gathered on the characteristics of this dye
with respect to the physical and chemical properties (98).
On the negative side, the fluorescence emission intensity of
fluorescein is heavily influenced by environmental factors
(e.g., pH), and the relatively broad emission spectrum often
overlaps with those of other fluorophores in dual and triple
labeling experiments (92,98,99).
Tetramethyl rhodamine (TMR) and the isothiocyanate
derivative (TRITC) are frequently employed in multiple
labeling investigations in widefield microscopy due to their
efficient excitation by the 546 nm spectral line from mercury arc-discharge lamps. The fluorochromes, which have
significant emission spectral overlap with fluorescein,
can be excited very effectively by the 543 nm line from
heliumneon lasers, but not by the 514 or 568 nanometer
lines from argon-ion and kryptonargon lasers (99). When
using krypton-based laser systems, Lissamine rhodamine
is a far better choice in this fluorochrome class due to the
absorption maximum at 575 nm and its spectral separation
from fluorescein. Also, the fluorescence emission intensity
of rhodamine derivatives is not as dependent upon strict
environmental conditions as that of fluorescein.
Several of the acridine dyes, first isolated in the nineteenth century, are useful as fluorescent probes in confocal
microscopy (93). The most widely utilized, acridine orange,
consists of the basic acridine nucleus with dimethylamino
substituents located at the 3 and 6 positions of the trinuclear ring system. In physiological pH ranges, the molecule is protonated at the heterocyclic nitrogen and exists
predominantly as a cationic species in solution. Acridine
orange binds strongly to DNA by intercalation of the
acridine nucleus between successive base pairs, and exhibits green fluorescence with a maximum wavelength of
530 nm (92,93,100). The probe also binds strongly to ribonucleic acid (RNA) or single-stranded deoxyribonucleic
467
acid (DNA), but has a longer wavelength fluorescence

maximum ( 640 nm; red) when bound to these macromolecules. In living cells, acridine orange diffuses across the
cell membrane (by virtue of the association constant for
protonation) and accumulates in the lysosomes and other
acidic vesicles. Similar to most acridines and related polynuclear nitrogen heterocycles, acridine orange has a relatively broad absorption spectrum, which enables the probe to
be used with several wavelengths from the argon-ion laser.
Another popular traditional probe that is useful in
confocal microscopy is the phenanthridine derivative, propidium iodide, first synthesized as an antitrypanosomal
agent along with the closely related ethidium bromide).
Propidium iodide binds to DNA in a manner similar to the
acridines (via intercalation) to produce orange-red fluorescence centered at 617 nm (101,102). The positively
charged fluorophore also has a high affinity for doublestranded RNA. Propidium has an absorption maximum at
536 nm, and can be excited by the 488 or 514-nm spectral
lines of an argon-ion (or kryptonargon) laser, or the
543 nm line from a green heliumneon laser. The dye is
often employed as a counterstain to highlight cell nuclei
during double or triple labeling of multiple intracellular
structures. Environmental factors can affect the fluorescence spectrum of propidium, especially when the dye is
used with mounting media containing glycerol. The structurally similar ethidium bromide, which also binds to DNA
by intercalation (102), produces more background staining,
and is therefore not as effective as propidium.
The DNA and chromatin can also be stained with dyes
that bind externally to the double helix. The most popular
fluorochromes in this category are 40 ,6-diamidino-2-phenylindole (DAPI) and the bis (benzimide) Hoechst dyes that
are designated by the numbers 33258, 33342, and 34580
(103106). These probes are quite water soluble and bind
externally to AT-rich base pair clusters in the minor groove
of double-stranded DNA with a dramatic increase in fluorescence intensity. Both dye classes can be stimulated by
the 351 nm spectral line of high power argon-ion lasers or
the 354 nm line from a heliumcadmium laser. Similar to
the acridines and phenanthridines, these fluorescent
probes are popular choices as a nuclear counterstain for
use in multicolor fluorescent labeling protocols. The vivid
blue fluorescence emission produces dramatic contrast
when coupled to green, yellow, and red probes in adjacent
cellular structures.
ALEXA FLUOR DYES

The dramatic advances in modern fluorophore technology
are exemplified by the Alexa Fluor dyes (90,107,108) introduced by Molecular Probes (Alexa Fluor is a registered
trademark of Molecular Probes). These sulfonated rhodamine derivatives exhibit higher quantum yields for more
intense fluorescence emission than spectrally similar
probes, and have several additional improved features,
including enhanced photostability, absorption spectra
matched to common laser lines, pH insensitivity, and a
high degree of water solubility. In fact, the resistance to
photobleaching of Alexa Fluor dyes is so dramatic (108)
468
that even when subjected to irradiation by high intensity

laser sources, fluorescence intensity remains stable for
relatively long periods of time in the absence of antifade
reagents. This feature enables the water soluble Alexa
Fluor probes to be readily utilized for both live-cell and
tissue section investigations, as well as in traditional fixed
preparations.
Alexa Fluor dyes are available in a broad range of
fluorescence excitation and emission wavelength maxima,
ranging from the UV and deep blue to the near-IR regions
(90). Alphanumeric names of the individual dyes are associated with the specific excitation laser or arc-discharge
lamp spectral lines for which the probes are intended. For
example, Alexa Fluor 488 is designed for excitation by the
blue 488 nm line of the argon or kryptonargon ion lasers,
while Alexa Fluor 568 is matched to the 568 nm spectral
line of the kryptonargon laser. Several of the Alexa Fluor
dyes are specifically designed for excitation by either the
blue diode laser (405 nm), the orange/yellow heliumneon
laser (594 nm), or the red heliumneon laser (633 nm).
Other Alexa Fluor dyes are intended for excitation with
traditional mercury arc-discharge lamps in the visible
(Alexa Fluor 546) or UV (Alexa Fluor 350, also useful with
high power argon-ion lasers), and solid-state red diode
lasers (Alexa Fluor 680). Because of the large number of
available excitation and emission wavelengths in the Alexa
Fluor series, multiple labeling experiments can often be
conducted exclusively with these dyes.
Alexa Fluor dyes are commercially available as reactive
intermediates in the form of maleimides, succinimidyl
esters, and hydrazides, as well as prepared cytoskeletal
probes (conjugated to phalloidin, G-actin, and rabbit skeletal muscle actin) and conjugates to lectin, dextrin, streptavidin, avidin, biocytin, and a wide variety of secondary
antibodies (90). In the latter forms, the Alexa Fluor fluorophores provide a broad palette of tools for investigations in
immunocytochemistry, neuroscience, and cellular biology.
The family of probes has also been extended into a series of
dyes having overlapping fluorescence emission maxima
targeted at sophisticated confocal microscopy detection
systems with spectral imaging and linear unmixing capabilities. For example, Alexa Fluor 488, Alexa Fluor 500,
and Alexa Fluor 514 are visually similar in color with
bright green fluorescence, but have spectrally distinct
emission profiles. In addition, the three fluorochromes
can be excited with the 488 or 514 nm spectral line from
an argon-ion laser and are easily detected with traditional
fluorescein filter combinations. In multispectral (xyl;
referred to as a lambda stack) confocal imaging experiments, optical separation software can be employed to
differentiate between the similar signals (3235). The overlapping emission spectra of Alexa Fluor 488, 500, and 514
are segregated into separate channels and differentiated
using pseudocolor techniques when the three fluorophores
are simultaneously combined in a triple label investigation.
CYANINE DYES
The popular family of cyanine dyes, Cy2, Cy3, Cy5, Cy7,
and their derivatives, are based on the partially saturated
indole nitrogen heterocyclic nucleus with two aromatic

units being connected via a polyalkene bridge of varying
carbon number (92,109). These probes exhibit fluorescence
excitation and emission profiles that are similar to many of
the traditional dyes, such as fluorescein and tetramethylrhodamine, but with enhanced water solubility, photostability, and higher quantum yields. Most of the cyanine dyes
are more environmentally stable than their traditional
counterparts, rendering their fluorescence emission intensity less sensitive to pH and organic mounting media.
In a manner similar to the Alexa Fluors, the excitation
wavelengths of the Cy series of synthetic dyes are tuned
specifically for use with common laser and arc-discharge
sources, and the fluorescence emission can be detected with
traditional filter combinations.
Marketed by a number of distributors, the cyanine dyes
are readily available as reactive dyes or fluorophores
coupled to a wide variety of secondary antibodies, dextrin,
streptavidin, and eggwhite avidin (110). The cyanine dyes
generally have broader absorption spectral regions than
members of the Alexa Fluor family, making them somewhat more versatile in the choice of laser excitation sources
for confocal microscopy (7). For example, using the 547 nm
spectral line from an argon-ion laser, Cy2 is about twice as
efficient in fluorescence emission as Alexa Fluor 488. In an
analogous manner, the 514 nm argon-ion laser line excites
Cy3 with a much higher efficiency than Alexa Fluor 546, a
spectrally similar probe. Emission profiles of the cyanine
dyes are comparable in spectral width to the Alexa Fluor
series.
Included in the cyanine dye series are the long-wavelength Cy5 derivatives, which are excited in the red region
(650 nm) and emit in the far-red (680 nm) wavelengths. The
Cy5 fluorophore is very efficiently excited by the 647 nm
spectral line of the kryptonargon laser, the 633 nm line of
the red heliumneon laser, or the 650 nm line of the red
diode laser, providing versatility in laser choice. Because
the emission spectral profile is significantly removed from
traditional fluorophores excited by UV and blue illumination, Cy5 is often utilized as a third fluorophore in triple
labeling experiments. However, similar to other probes
with fluorescence emission in the far-red spectral region,
Cy5 is not visible to the human eye and can only be
detected electronically (using a specialized CCD camera
system or photomultiplier). Therefore, the probe is seldom
used in conventional widefield fluorescence experiments.
FLUORESCENT ENVIRONMENTAL PROBES

Fluorophores designed to probe the internal environment
of living cells have been widely examined by a number of
investigators, and many hundreds have been developed to
monitor such effects as localized concentrations of alkali
and alkaline earth metals, heavy metals (employed biochemically as enzyme cofactors), inorganic ions, thiols, and
sulfides, nitrite, as well as pH, solvent polarity, and membrane potential (7,9093,111,112). Originally, the experiments in this arena were focused on changes in the
wavelength and/or intensity of absorption and emission
spectra exhibited by fluorophores upon binding calcium
ions in order to measure intracellular flux densities. These

probes bind to the target ion with a high degree of specificity to produce the measured response and are often
referred to as spectrally sensitive indicators. Ionic concentration changes are determined by the application of
optical ratio signal analysis to monitor the association
equilibrium between the ion and its host. The concentration values derived from this technique are largely
independent of instrumental variations and probe concentration fluctuations due to photobleaching, loading parameters, and cell retention. In the past few years, a number
of new agents have been developed that bind specific ions or
respond with measurable features to other environmental
conditions (7,90).
Calcium is a metabolically important ion that plays a
vital role in cellular response to many forms of external
stimuli (113). Because transient fluctuations in calcium ion
concentration are typically involved when cells undergo a
response, fluorophores must be designed to measure not
only localized concentrations within segregated compartments, but should also produce quantitative changes when
flux density waves progress throughout the entire cytoplasm. Many of the synthetic molecules designed to measure calcium levels are based on the nonfluorescent
chelation agents EGTA and BAPTA, which have been used
for years to sequester calcium ions in buffer solutions
(7,114,115). Two of the most common calcium probes are
the ratiometric indicators fura-2 and indo-1, but these
fluorophores are not particularly useful in confocal microscopy (7,116). The dyes are excited by UV light and exhibit
a shift in the excitation or emission spectrum with the
formation of isosbestic points when binding calcium. However, the optical aberrations associated with UV imaging,
limited specimen penetration depths, and the expense of
ultraviolet lasers have limited the utility of these probes in
confocal microscopy.
Fluorophores that respond in the visible range to calcium ion fluxes are, unfortunately, not ratiometric indicators and do not exhibit a wavelength shift (typical of fura-2
and indo-1) upon binding, although they do undergo an
increase or decrease in fluorescence intensity. The best
example is fluo-3, a complex xanthene derivative, which
undergoes a dramatic increase in fluorescence emission at
525 nm (green) when excited by the 488 nm spectral line of
an argon-ion or kryptonargon laser (7,117). Because isosbestic points are not present to assure the absence of concentration fluctuations, it is impossible to determine whether
spectral changes are due to complex formation or a variation
in concentration with fluo-3 and similar fluorophores.
To overcome the problems associated with using visible
light probes lacking wavelength shifts (and isosbestic
points), several of these dyes are often utilized in combination for calcium measurements in confocal microscopy
(118). Fura red, a multinuclear imidazole and benzofuran
heterocycle, exhibits a decrease in fluorescence at 650 nm
when binding calcium. A ratiometric response to calcium
ion fluxes can be obtained when a mixture of fluo-3 and fura
red is excited at 488 nm and fluorescence is measured at the
emission maxima (525 and 650 nm, respectively) of the two
probes. Because the emission intensity of fluo-3 increases
monotonically while that of fura red simultaneously
469
decreases, an isosbestic point is obtained when the dye

concentrations are constant within the localized area
being investigated. Another benefit of using these probes
together is the ability to measure fluorescence intensity
fluctuations with a standard FITC/Texas red interference
filter combination.
Quantitative measurements of ions other than calcium,
such as magnesium, sodium, potassium, and zinc, are
conducted in an analogous manner using similar fluorophores (7,90,92). One of the most popular probes for magnesium, mag-fura-2 (structurally similar to fura red), is
also excited in the ultraviolet range and presents the same
problems in confocal microscopy as fura-2 and indo-1.
Fluorophores excited in the visible light region are becoming available for the analysis of many monovalent and
divalent cations that exist at varying concentrations in
the cellular matrix. Several synthetic organic probes have
also been developed for monitoring the concentration of
simple and complex anions.
Important fluorescence monitors for intracellular pH
include a pyrene derivative known as HPTS or pyranine,
the fluorescein derivative, BCECF, and another substituted xanthene termed carboxy SNARF-1 (90,119122).
Because many common fluorophores are sensitive to pH
in the surrounding medium, changes in fluorescence intensity that are often attributed to biological interactions may
actually occur as a result of protonation. In the physiological
pH range (pH 6.87.4), the probes mentioned above are
useful for dual-wavelength ratiometric measurements and
differ only in dye loading parameters. Simultaneous measurements of calcium ion concentration and pH can often be
accomplished by combining a pH indicator, such as SNARF1, with a calcium ion indicator (e.g., fura-2). Other probes
have been developed for pH measurements in subcellular
compartments, such as the lysosomes, as described below.
ORGANELLE PROBES
Fluorophores targeted at specific intracellular organelles,
such as the mitochondria, lysosomes, Golgi apparatus, and
endoplasmic reticulum, are useful for monitoring a variety
of biological processes in living cells using confocal microscopy (7,90,92). In general, organelle probes consist of a
fluorochrome nucleus attached to a target-specific moiety
that assists in localizing the fluorophore through covalent,
electrostatic, hydrophobic, or similar types of bonds. Many
of the fluorescent probes designed for selecting organelles
are able to permeate or sequester within the cell membrane
(and therefore, are useful in living cells), while others must
be installed using monoclonal antibodies with traditional
immunocytochemistry techniques. In living cells, organelle
probes are useful for investigating transport, respiration,
mitosis, apoptosis, protein degradation, acidic compartments, and membrane phenomena. Cell impermeant fluorophore applications include nuclear functions, cytoskeletal
structure, organelle detection, and probes for membrane
integrity. In many cases, living cells that have been labeled
with permeant probes can subsequently be fixed and counterstained with additional fluorophores in multicolor labeling experiments.
470
Mitochondrial probes are among the most useful fluorophores for investigating cellular respiration and are often
employed along with other dyes in multiple labeling investigations. The traditional probes, rhodamine 123 and tetramethylrosamine, are rapidly lost when cells are fixed
and have largely been supplanted by newer, more specific,
fluorophores developed by Molecular Probes (90,123,124).
These include the popular MitoTracker and MitoFluor
series of structurally diverse xanthene, benzoxazole,
indole, and benzimidazole heterocycles that are available
in a variety of excitation and emission spectral profiles. The
mechanism of action varies for each of the probes in this
series, ranging from covalent attachment to oxidation
within respiring mitochondrial membranes.
MitoTracker dyes are retained quite well after cell
fixation in formaldehyde and can often withstand lipophilic
permeabilizing agents (123). In contrast, the MitoFluor
probes are designed specifically for actively respiring cells
and are not suitable for fixation and counterstaining procedures (90). Another popular mitochondrial probe,
entitled JC-1, is useful as an indicator of membrane potential and in multiple staining experiments with fixed cells
(125). This carbocyanine dye exhibits green fluorescence at
low concentrations, but can undergo intramolecular association within active mitochondria to produce a shift in
emission to longer (red) wavelengths. The change in emission wavelength is useful in determining the ratio of active
to nonactive mitochondria in living cells.
In general, weakly basic amines that are able to
pass through membranes are the ideal candidates for
investigating biosynthesis and pathogenesis in lysosomes
(9092,112). Traditional lysosomal probes include the nonspecific phenazine and acridine derivatives neutral red and
acridine orange, which are accumulated in the acidic vesicles upon being protonated (92,93). Fluorescently labeled
latex beads and macromolecules, such as dextran, can also
be accumulated in lysosomes by endocytosis for a variety of
experiments. However, the most useful tools for investigating lysosomal properties with confocal microscopy are
the LysoTracker and LysoSensor dyes developed by Molecular Probes (90,92,126). These structurally diverse agents
contain heterocyclic and aliphatic nitrogen moieties that
modulate transport of the dyes into the lysosomes of living
cells for both short- and long-term studies. The LysoTracker
probes, which are available in a variety of excitation and
emission wavelengths (91), have high selectivity for acidic
organelles and are capable of labeling cells at nanomolar
concentrations. Several of the dyes are retained quite well
after fixing and permeabilization of cells. In contrast, the
LysoSensor fluorophores are designed for studying dynamic
aspects of lysosome function in living cells. Fluorescence
intensity dramatically increases in the LysoSensor series
upon protonation, making these dyes useful as pH indicators (91). A variety of Golgi apparatus specific monoclonal
antibodies have also been developed for use in immunocytochemistry assays (90,127129).
Proteins and lipids are sorted and processed in the Golgi
apparatus, which is typically stained with fluorescent
derivatives of ceramides and sphingolipids (130). These
agents are highly lipophilic, and are therefore useful as
markers for the study of lipid transport and metabolism in
live cells. Several of the most useful fluorophores for Golgi

apparatus contain the complex heterocyclic BODIPY
nucleus developed by Molecular Probes (90,92,131). When
coupled to sphingolipids, the BODIPY fluorophore is highly
selective and exhibits a tolerance for photobleaching that is
far superior to many other dyes. In addition, the emission
spectrum is dependent upon concentration (shifting from
green to red at higher concentrations), making the probes
useful for locating and identifying intracellular structures
that accumulate large quantities of lipids. During live-cell
experiments, fluorescent lipid probes can undergo metabolism to derivatives that may bind to other subcellular
features, a factor that can often complicate the analysis of
experimental data.
The most popular traditional probes for endoplasmic
reticulum fluorescence analysis are the carbocyanine and
xanthene dyes, DiOC (6) and several rhodamine derivatives, respectively (90,92). These dyes must be used with
caution, however, because they can also accumulate in the
mitochondria, Golgi apparatus, and other intracellular
lipophilic regions. Newer, more photostable, probes have
been developed for selective staining of the endoplasmic
reticulum by several manufacturers. In particular, oxazole
members of the Dapoxyl family produced by Molecular
Probes are excellent agents for selective labeling of the
endoplasmic reticulum in living cells, either alone or in
combination with other dyes (90). These probes are
retained after fixation with formaldehyde, but can be lost
with permeabilizing detergents. Another useful probe is
Brefeldin A (131), a stereochemically complex fungal metabolite that serves as an inhibitor of protein trafficking out
of the endoplasmic reticulum. Finally, similar to other
organelles, monoclonal antibodies (127129) have been
developed that target the endoplasmic reticulum in fixed
cells for immunocytochemistry investigations.
QUANTUM DOTS
Nanometer-sized crystals of purified semiconductors
known as quantum dots are emerging as a potentially
useful fluorescent labeling agent for living and fixed cells
in both traditional widefield and laser scanning confocal
fluorescence microscopy (132136). Recently introduced
techniques enable the purified tiny semiconductor crystals
to be coated with a hydrophilic polymer shell and conjugated to antibodies or other biologically active peptides and
carbohydrates for application in many of the classical
immunocytochemistry protocols (Fig. 15). These probes
have significant benefits over organic dyes and fluorescent
proteins, including long-term photostability, high fluorescence intensity levels, and multiple colors with singlewavelength excitation for all emission profiles (136).
Quantum dots produce illumination in a manner similar
to the well-known semiconductor light emitting diodes, but
are activated by absorption of a photon rather than an
electrical stimulus. The absorbed photon creates an electron-hole pair that quickly recombines with the concurrent
emission of a photon having lower energy. The most useful
semiconductor discovered thus far for producing biological
quantum dots is cadmium selenide (CdSe), a material in
Figure 15. Anatomy and spectral profiles of quantum dot

conjugates. The cadmium selenide core is encapsulated with
zinc sulfide, and then a polymer coating is applied followed by a
hydrophilic exterior to which the biological conjugate is attached
(left). The absorption profile displays a shoulder at 400 nm, while
the emission spectra all feature similar symmetrical profiles.
which the energy of the emitted photons is a function of the

physical size of the nanocrystal particles. Thus, quantum
dots having sizes that differ only by tenths of a nanometer
emit different wavelengths of light, with the smaller sizes
emitting shorter wavelengths, and vice versa.
Unlike typical organic fluorophores or fluorescent proteins, which display highly defined spectral profiles, quantum dots have an absorption spectrum that increases
steadily with decreasing wavelength (Fig. 15). Also, in
contrast, the fluorescence emission intensity is confined
to a symmetrical peak with a maximum wavelength that is
dependent on the dot size, but independent of the excitation wavelength (135). As a result, the same emission
profile is observed regardless of whether the quantum
dot is excited at 300, 400, 500, or 600 nm, but the fluorescence intensity increases dramatically at shorter excitation
wavelengths. For example, the extinction coefficient for a
typical quantum dot conjugate that emits in the orange
region (605 nm) is approximately five-fold higher when the
semiconductor is excited at 400 versus 600 nm. The fwhm
value for a typical quantum dot conjugate is 30 nm (135),
and the spectral profile is not skewed towards the longer
wavelengths (having higher intensity tails), such is the case
with most organic fluorochromes. The narrow emission
profile enables several quantum dot conjugates to be simultaneously observed with a minimal level of bleed through.
For biological applications, a relatively uniform population of cadmium selenide crystals is covered with a surrounding semiconductor shell composed of zinc sulfide to
improve the optical properties. Next, the core material is
coated with a polymeric film and other ligands to decrease
hydrophobicity and to improve the attachment efficiency of
conjugated macromolecules. The final product is a biologically active particle that ranges in size from 10 to 15 nm,
somewhere in the vicinity of a large protein (133). Quantum dot conjugates are solubilized as a colloidal suspension
in common biological buffers and may be incorporated into
existing labeling protocols in place of classical staining
reagents (such as organic fluorochrome-labeled secondary
antibodies).
In confocal microscopy, quantum dots are excited with
varying degrees of efficiency by most of the spectral lines
produced by the common laser systems, including the
471
argon-ion, heliumcadmium, kryptonargon, and the

green heliumneon. Particularly effective at exciting quantum dots in the UV and violet regions are the new blue
diode and diode-pumped solid-state lasers that have prominent spectral lines at 442 nm and below (135,136). The
405 nm blue diode laser is an economical excitation source
that is very effective for use with quantum dots due to their
high extinction coefficient at this wavelength. Another
advantage of using these fluorophores in confocal microscopy is the ability to stimulate multiple quantum dot sizes
(and spectral colors) in the same specimen with a single
excitation wavelength, making these probes excellent candidates for multiple labeling experiments (137).
The exceptional photostability of quantum dot conjugates
is of great advantage in confocal microscopy when optical
sections are being collected. Unlike the case of organic
fluorophores, labeled structures situated away from the
focal plane do not suffer from excessive photobleaching
during repeated raster scanning of the specimen and yield
more accurate 3D volume models. In widefield fluorescence
microscopy, quantum dot conjugates are available for use
with conventional dye-optimized filter combinations that
are standard equipment on many microscopes. Excitation
can be further enhanced by substituting a shortpass filter
for the bandpass filter that accompanies most filter sets,
thus optimizing the amount of lamp energy that can be
utilized to excite the quantum dots. Several of the custom
fluorescence filter manufacturers offer combinations specifically designed to be used with quantum dot conjugates.
FLUORESCENT PROTEINS
Over the past few years, the discovery and development of
naturally occurring fluorescent proteins and mutated derivatives have rapidly advanced to center stage in the investigation of a wide spectrum of intracellular processes in
living organisms (75,78,80). These biological probes have
provided scientists with the ability to visualize, monitor,
and track individual molecules with high spatial and temporal resolution in both steady-state and kinetic experiments. A variety of marine organisms have been the source
of >100 fluorescent proteins and their analogs, which arm
the investigator with a balanced palette of noninvasive
biological probes for single, dual, and multispectral fluorescence analysis (75). Among the advantages of fluorescent
proteins over the traditional organic and new semiconductor probes described above is their response to a wider
variety of biological events and signals. Coupled with the
ability to specifically target fluorescent probes in subcellular compartments, the extremely low or absent photodynamic toxicity, and the widespread compatibility with
tissues and intact organisms, these biological macromolecules offer an exciting new frontier in live-cell imaging.
The first member of this series to be discovered, GFP,
was isolated from the North Atlantic jellyfish, Aequorea
Victoria, and found to exhibit a high degree of fluorescence
without the aid of additional substrates or coenzymes (138
142). In native green fluorescent protein, the fluorescent
moiety is a tripeptide derivative of serine, tyrosine, and
glycine that requires molecular oxygen for activation, but
no additional cofactors or enzymes (143). Subsequent
472
investigations revealed that the GFP gene could be

expressed in other organisms, including mammals, to yield
fully functional analogs that display no adverse biological
effects (144). In fact, fluorescent proteins can be fused to
virtually any protein in living cells using recombinant
complementary DNA cloning technology, and the resulting
fusion protein gene product expressed in cell lines adapted
to standard tissue culture methodology. Lack of a need for
cell-specific activation cofactors renders the fluorescent
proteins much more useful as generalized probes than
other biological macromolecules, such as the phycobiliproteins, which require insertion of accessory pigments in
order to produce fluorescence.
Mutagenesis experiments with green fluorescent protein have produced a large number of variants with
improved folding and expression characteristics, which
have eliminated wild-type dimerization artifacts and fine
tuned the absorption and fluorescence properties. One of
the earliest variants, known as enhanced green fluorescence protein (EGFP), contains codon substitutions (commonly referred to as the S65T mutation) that alleviates the
temperature sensitivity and increases the efficiency of GFP
expression in mammalian cells (145). Proteins fused with
EGFP can be observed at low light intensities for long time
periods with minimal photobleaching. Enhanced green
fluorescent protein fusion products are optimally excited
by the 488 nm spectral line from argon and kryptonargon
ion lasers in confocal microscopy. This provides an excellent biological probe and instrument combination for examining intracellular protein pathways along with the
structural dynamics of organelles and the cytoskeleton.
Additional mutation studies have uncovered GFP variants that exhibit a variety of absorption and emission
characteristics across the entire visible spectral region,
which have enabled researchers to develop probe combinations for simultaneous observation of two or more distinct
fluorescent proteins in a single organism (see the spectral
profiles in Fig. 16). Early investigations yielded the blue
fluorescent protein (BFP) and cyan fluorescent protein
(CFP) mutants from simple amino acid substitutions that
shifted the absorption and emission spectral profiles of
wild-type GFP to lower wavelength regions (146148).
Used in combination with GFP, these derivatives are useful in resonance energy transfer (FRET) experiments and
other investigations that rely on multicolor fluorescence
imaging (73). Blue fluorescent protein can be efficiently
excited with the 354 nm line from a high power argon laser,
while the more useful cyan derivative is excited by a
number of violet and blue laser lines, including the
Figure 16. Fluorescent spectral profiles, plotted as normalized

absorption or emission as a function of wavelength, for fluorescent proteins emitting in the blue to orange-red regions of the
visible spectrum. Each profile is identified with a colored bullet in
(a), which illustrates excitation spectra. (b) The emission spectra
for the proteins according to the legend in (a).
405 nm blue diode, the 442 nm heliumcadmium spectral

line, and the 457 nm line from the standard argon-ion laser.
Another popular fluorescent protein derivative, the yellow fluorescent protein (YFP), was designed on the basis of
the GFP crystalline structural analysis to red-shift the
absorption and emission spectra (148). Yellow fluorescent
protein is optimally excited by the 514 nm spectral line of
the argon-ion laser, and provides more intense emission
than enhanced green fluorescent protein, but is more
sensitive to low pH and high halogen ion concentrations.
The enhanced yellow fluorescent protein derivative
(EYFP) is useful with the 514 argon-ion laser line, but
can also be excited with relatively high efficiency by the
488 nm line from argon and kryptonargon lasers. Both of
these fluorescent protein derivatives have been widely
applied to proteinprotein FRET investigations in combination with CFP, and in addition, have proven useful in
studies involving multiprotein trafficking.
Attempts to shift the absorption and emission spectra of
Aequorea Victoria fluorescent proteins to wavelengths in
the orange and red regions of the spectrum have met with
little success. However, fluorescent proteins from other
marine species have enabled investigators to extend the
available spectral regions to well within the red wavelength range. The DsRed fluorescent protein and its
derivatives, originally isolated from the sea anemone
Discosoma striata, are currently the most popular analogs
for fluorescence analysis in the 575650 nm region (149).
Another protein, HcRed from the Heteractis crispa purple
anemone, is also a promising candidate for investigations
in the longer wavelengths of the visible spectrum (150).
Newly developed photoactivation fluorescent proteins,
including photoactivatable green fluorescent protein
(PA-GFP;74), Kaede (76), and kindling fluorescent protein
1 (KFP1; 151), exhibit dramatic improvements over GFP
(up to several 1000-fold) in fluorescence intensity when
stimulated by violet laser illumination. These probes
should prove useful in fluorescence confocal studies involving selective irradiation of specific target regions and the
subsequent kinetic analysis of diffusional mobility and
compartmental residency time of fusion proteins.
QUENCHING AND PHOTOBLEACHING

The consequences of quenching and photobleaching are
suffered in practically all forms of fluorescence microscopy, and result in an effective reduction in the levels
of emission (152,153). These artifacts should be of primary
consideration when designing and executing fluorescence

investigations. The two phenomena are distinct in that
quenching is often reversible whereas photobleaching is
not (154). Quenching arises from a variety of competing
processes that induce nonradiative relaxation (without
photon emission) of excited-state electrons to the ground
state, which may be either intramolecular or intermolecular in nature. Because nonradiative transition pathways compete with the fluorescence relaxation, they
usually dramatically lower or, in some cases, completely
eliminate emission. Most quenching processes act to
reduce the excited state lifetime and the quantum yield
of the affected fluorophore.
A common example of quenching is observed with the
collision of an excited state fluorophore and another (nonfluorescent) molecule in solution, resulting in deactivation of
the fluorophore and return to the ground state. In most
cases, neither of the molecules is chemically altered in the
collisional quenching process. A wide variety of simple
elements and compounds behave as collisional quenching
agents, including oxygen, halogens, amines, and many electron-deficient organic molecules (154). Collisional quenching can reveal the presence of localized quencher molecules
or moieties, which via diffusion or conformational change,
may collide with the fluorophore during the excited state
lifetime. The mechanisms for collisional quenching include
electron transfer, spinorbit coupling, and intersystem
crossing to the excited triplet state (154,155). Other terms
that are often utilized interchangeably with collisional
quenching are internal conversion and dynamic quenching.
A second type of quenching mechanism, termed static
or complex quenching, arises from nonfluorescent complexes formed between the quencher and fluorophore that
serve to limit absorption by reducing the population of
active, excitable molecules (154,156). This effect occurs
when the fluorescent species forms a reversible complex
with the quencher molecule in the ground state, and does
not rely on diffusion or molecular collisions. In static
quenching, fluorescence emission is reduced without altering the excited state lifetime. A fluorophore in the excited
state can also be quenched by a dipolar resonance energy
transfer mechanism when in close proximity with an
acceptor molecule to which the excited-state energy can
be transferred nonradiatively. In some cases, quenching
can occur through non molecular mechanisms, such as
attenuation of incident light by an absorbing species
(including the chromophore itself).
In contrast to quenching, photobleaching (also
termed fading) occurs when a fluorophore permanently
loses the ability to fluoresce due to photon-induced chemical damage and covalent modification (153156). Upon
transition from an excited singlet state to the excited
triplet state, fluorophores may interact with another molecule to produce irreversible covalent modifications. The
triplet state is relatively long lived with respect to the
singlet state, thus allowing excited molecules a much
longer timeframe to undergo chemical reactions with components in the environment (155). The average number of
excitation and emission cycles that occur for a particular
fluorophore before photobleaching is dependent on the
molecular structure and the local environment (154,156).
473
Some fluorophores bleach quickly after emitting only a few

photons, while others that are more robust can undergo
thousands or even millions of cycles before bleaching.
Figure 17 presents a typical example of photobleaching
(fading) observed in a series of digital images captured at
different time points for a multiply stained culture of
normal Tahr ovary (HJ1.Ov line) fibroblast cells. The
nuclei were stained with DAPI (blue fluorescence), while
the mitochondria and actin cytoskeleton were stained with
MitoTracker Red CMXRos (red fluorescence) and an Alexa
Fluor phalloidin derivative (Alexa Fluor 488; green fluorescence), respectively. Time points were taken in 2 min
intervals using a fluorescence filter combination with
bandwidths tuned to excite the three fluorophores simultaneously while also recording the combined emission
signals. Note that all three fluorophores have a relatively
high intensity in Fig. 17a, but the DAPI (blue) intensity
starts to drop rapidly at two min and is almost completely
gone at six min (Fig. 17f). The mitochondrial and actin
stains are more resistant to photobleaching, but the intensity of both drops dramatically over the course of the timed
sequence (10 min).
An important class of photobleaching events is represented by events that are photodynamic, meaning they
involve the interaction of the fluorophore with a combination of light and oxygen (157161). Reactions between
fluorophores and molecular oxygen permanently destroy
fluorescence and yield a free-radical singlet oxygen species
that can chemically modify other molecules in living cells.
The amount of photobleaching due to photodynamic events
is a function of the molecular oxygen concentration and
the proximal distance between the fluorophore, oxygen
molecules, and other cellular components. Photobleaching
can be reduced by limiting the exposure time of
Figure 17. Photobleaching in multiply stained specimens. Normal

Tahr ovary fibroblast cells were stained with MitoTracker Red
CMXRos (mitochondria; red fluorescence), Alexa Fluor 488
conjugated to phalloidin (actin; green fluorescence), and
subsequently counterstained with DAPI (nuclei; blue fluorescence). Time points were taken in two-minute intervals over a
10 min period using a fluorescence filter combination with
bandwidths tuned to excite the three fluorophores simultaneously
while also recording the combined emission signals. (af) Time 0, 2,
4, 6, 8, 10 min, respectively.
474
fluorophores to illumination or by lowering the excitation

energy. However, these techniques also reduce the measurable fluorescence signal. In many cases, solutions of
fluorophores or cell suspensions can be deoxygenated, but
this is not feasible for living cells and tissues. Perhaps the
best protection against photobleaching is to limit exposure
of the fluorochrome to intense illumination (using neutral
density filters) coupled with the judicious use of commercially available antifade reagents that can be added to the
mounting solution or cell culture medium (153).
Under certain circumstances, the photobleaching effect
can also be utilized to obtain specific information that
would not otherwise be available. For example, in FRAP
experiments, fluorophores within a target region are intentionally bleached with excessive levels of irradiation (82).
As new fluorophore molecules diffuse into the bleached
region of the specimen (recovery), the fluorescence emission intensity is monitored to determine the lateral diffusion rates of the target fluorophore. In this manner, the
translational mobility of fluorescently labeled molecules
can be ascertained within a very small (25 mm) region of a
single cell or section of living tissue.
Although the subset of fluorophores that are advantageous in confocal microscopy is rapidly growing, many of
the traditional probes that have been useful for years in
widefield applications are still of little utility when constrained by fixed-wavelength laser spectral lines. Many of
the limitations surrounding the use of fluorophores excited
in the ultraviolet region will be eliminated with the introduction of advanced objectives designed to reduce aberration coupled to the gradual introduction of low cost, high
power diode laser systems with spectral lines in these
shorter wavelengths. The 405 nm blue diode laser is a
rather cheap alternative to more expensive ion and Noble
gas based ultraviolet lasers, and is rapidly becoming
available for most confocal microscope systems. Helium
neon lasers with spectral lines in the yellow and orange
region have rendered some fluorophores useful that were
previously limited to widefield applications. In addition,
new diode-pumped solid-state lasers are being introduced
with emission wavelengths in the UV, violet, and blue
regions.
Continued advances in fluorophore design, dual-laser
scanning, multispectral imaging, endoscopic instruments,
and spinning disk applications will also be important in the
coming years. The persistent problem of emission crossover
due to spectral overlap, which occurs with many synthetic
probes and fluorescent proteins in multicolor investigations, benefits significantly from spectral analysis and
deconvolution of lambda stacks. Combined, these advances
and will dramatically improve the collection and analysis of
data obtained from complex fluorescence experiments in
live-cell imaging.
BIBLIOGRAPHY
1. Pawley JB, editor. Handbook of Biological Confocal Microscopy. New York: Plenum Press; 1995.
2. Paddock SW, editor. Confocal Microscopy: Methods and Protocols. Totowa (NJ): Humana Press; 1999.
3. Diaspro A, editor. Confocal and Two-Photon Microscopy:

Foundations, Applications, and Advances. New York:
Wiley-Liss; 2002.
4. Matsumoto B, editor. Cell Biological Applications of Confocal
Microscopy. Methods in Cell Biology, Vol. 70. New York:
Academic Press; 2002.
5. Sheppard CJR, Shotton DM. Confocal Laser Scanning Microscopy. Oxford (UK): BIOS Scientific Publishers; 1997.
6. Mu ller M. Introduction to Confocal Fluorescence Microscopy.
Maastricht, The Netherlands: Shaker; 2002.
7. Hibbs AR. Confocal Microscopy for Biologists. New York:
Kluwer Academic; 2004.
8. Conn PM. Confocal Microscopy. Methods in Enzymology, Vol.
307. New York: Academic Press; 1999.
9. Corle TR, Kino GS. Confocal Scanning Optical Microscopy and
Related Imaging Systems. New York: Academic Press; 1996.
10. Wilson T, editor. Confocal Microscopy. New York: Academic
Press; 1990.
11. Gu M. Principles of Three-Dimensional Imaging in Confocal
Microscopes. New Jersey: World Scientific; 1996.
12. Masters BR, editor. Selected Papers on Confocal Microscopy.
SPIE Milestone Series, Vol. MS 131. Bellingham (WA): SPIE
Optical Engineering Press; 1996.
13. Mason WT. Fluorescent and Luminescent Probes for Biological Activity. New York: Academic Press; 1999.
14. Peterman EJG, Sosa H, Moerner WE. Single-Molecule Fluorescence Spectroscopy and Microscopy of Biomolecular
Motors. Ann Rev Phys Chem 2004;55:7996.
15. Goldman RD, Spector DL. Live Cell Imaging: A Laboratory
Manual. New York: Cold Spring Harbor Press; 2005.
16. Minsky M. Microscopy Apparatus. US Patent 3,013,467. 1961.
17. Minsky M. Memoir on Inventing the Confocal Scanning
Microscopy. Scanning 1988;10:128138.
18. Egger MD, Petran M. New Reflected-Light Microscope for
Viewing Unstained Brain and Ganglion Cells. Science
1967;157:305307.
19. Davidovits P, Egger MD. Photomicrography of Corneal
Endothelial Cells in vivo. Nature (London) 1973;244:366
367.
20. Amos WB, White JG. How the Confocal Laser Scanning Microscope entered Biological Research. Biol Cell 2003;95:335342.
21. Brakenhoff GJ, Blom P, Barends P. Confocal Scanning Light
Microscopy with High Aperture Immersion Lenses. J Micros
1979;117:219232.
22. Sheppard CJR, Wilson T. Effect of Spherical Aberration on
the Imaging Properties of Scanning Optical Microscopes.
Appl Opt 1979;18:1058.
23. Hamilton DK, Wilson T. Scanning Optical Microscopy by
Objective Lens Scanning. J Phys E: Sci Instr 1986;19:5254.
24. Spring KR, Inoue S. Video Microscopy: The Fundamentals.
25. Wilson T. Optical Sectioning in Confocal Fluorescence Microscopes. J Micros 1989;154:143156.
26. Lichtmann JW. Confocal Microscopy. Sci Am Aug. 1994; 4045.
27. White JG, Amos WB, Fordham M. An Evaluation of Confocal
versus Conventional Imaging of Biological Structures by
Fluorescence Light Microscopy. J Cell Biol 1987;105:4148.
28. Swedlow JR, et al. Measuring Tubulin Content in Toxoplasma gondii: A Comparison of Laser-Scanning Confocal
and Wide-Field Fluorescence Microscopy. Proc Natl Acad
Sci USA 2002;99:20142019.
29. Stelzer EHK. Practical Limits to Resolution in Fluorescence
Light Microscopy. In: Yuste R, Lanni F, Konnerth A, editors.
Imaging Neurons: A Laboratory Manual. New York: Cold
Spring Harbor Press; 2000. pp 12.112.9.
30. Rost FWD. Fluorescence Microscopy. Vol. 1. New York: Cambridge University Press; 1992.
31. Murray J. Confocal Microscopy, Deconvolution, and Structured Illumination Methods. In: Goldman RD, Spector DL,
editors. Live Cell Imaging: A Laboratory Manual. New York:
Cold Spring Harbor Press; 2005. pp 239280.
32. Dickinson ME, et al. Multi-Spectral Imaging and Linear
Unmixing Add a Whole New Dimension to Laser Scanning
Fluorescence Microscopy. Biotechniques 2001;31:12721278.
33. Zimmermann T, Rietdorf J, Pepperkok R. Spectral Imaging
and its Applications in Live Cell Microscopy. FEBS Lett
2003;546:8792.
34. Lansford R, Bearman G, Fraser SE. Resolution of Multiple
Green Fluorescent Protein Variants and Dyes using TwoPhoton Microscopy and Imaging Spectroscopy. J Biomed Opt
2001;6:311318.
35. Hiraoka Y, Shimi T, Haraguchi T. Multispectral Imaging
Fluorescence Microscopy for Living Cells. Cell Struct Funct
2002;27:367374.
36. Gu Y, Di WL, Kelsell DP, Zicha D. Quantitative Fluorescence
Energy Transfer (FRET) Measurement with Acceptor Photobleaching and Spectral Unmixing. J Microsc 2004;215:162
173.
37. Ford BK, et al. Computed Tomography-Based Spectral Imaging for Fluorescence Microscopy. Biophys J 2001;80:986
993.
38. Bach H, Renn A, Wild UP. Spectral Imaging of Single Molecules. Single Mol 2000;1:7377.
39. Wilson T, Carlini AR. Three-Dimensional Imaging in Confocal Imaging Systems with Finite Sized Detectors. J Microsc
1988;149:5166.
40. Murphy DB. Fundamentals of Light Microscopy and Electronic Imaging. New York: Wiley-Liss; 2001.
41. Wright SJ, Wright DJ. Introduction to Confocal Microscopy.
In: Matsumoto B, editor. Cell Biological Applications of Confocal Microscopy. Methods in Cell Biology, Vol. 70. New York:
Academic Press; 2002. pp 185.
42. Webb RH. Confocal Optical Microscopy. Rep Prog Phys
1996;59:427471.
43. Wilhelm S, Grobler B, Gluch M, Hartmut H. Confocal Laser
Scanning Microscopy: Optical Image Formation and Electronic Signal Processing. Jena, Germany: Carl Zeiss Advanced
Imaging Microscopy; 2003.
44. Ichihara A, et al. High-Speed Confocal Fluorescence Microscopy using a Nipkow Scanner with Microlenses for 3-D
Imaging of Fluorescent Molecules in Real-Time. Bioimages
1996;4:5762.
45. Inoue S, Inoue T. Direct-View High-Speed Confocal ScannerThe CSU-10. In: Matsumoto B, editor. Cell Biological
Applications of Confocal Microscopy. Methods in Cell Biology,
Vol. 70. New York: Academic Press; 2002. p 88128.
46. Nakano A. Spinning-Disk Confocal Microscopy A CuttingEdge Tool for Imaging of Membrane Traffic. Cell Struct Funct
2002;27:349355.
47. Chong FK, et al. Optimization of Spinning Disk Confocal
Microscopy: Synchronization with the Ultra-Sensitive
EMCCD. In: Conchello JA, Cogswell CJ, Wilson T, editors.
Three-Dimensional and Multidimensional Microscopy: Image
Acquisition and Processing XI. Proc. SPIE. 2004; 5324: 6576.
48. Sandison D, Webb W. Background Rejection and Signal-toNoise Optimization in the Confocal and Alternative Fluorescence Microscopes. Appl Op 1994;33:603610.
49. Centonze VE, White JG. Multiphoton Excitation Provides
Optical Sections from Deeper within Scattering Specimens
than Confocal Imaging. Biophys J 1998;75:2015.
50. Boccacci P, Bertero M. Image-Restoration Methods: Basics
and Algorithms. In: Diaspro A, editor. Confocal and TwoPhoton Microscopy: Foundations, Applications, and Advances.
New York: Wiley-Liss; 2002. pp 253269.
475
51. Verveer PJ, Gemkow MJ, Jovin TM. A Comparison of Image

Restoration Approaches Applied to Three-Dimensional Confocal and Wide-Field Fluorescence Microscopy. J Micros
1998;193:5061.
52. Conchello JA, Hansen EW. Enhanced 3-D Reconstruction
from Confocal Scanning Microscope Images. 1: Deterministic
and Maximum Likelihood Reconstructions. Appl Op 1990;29:
37953804.
53. Al-Kofahi O, et al. Algorithms for Accurate 3D Registration of
Neuronal Images Acquired by Confocal Scanning Laser
Microscopy. J Micros 2003;211:818.
54. Conchello JA, et al., editors. Three-Dimensional and Multidimensional Microscopy: Image Acquisition and Processing.
Vol. IXII, Bellingham (WA): SPIE International Society for
Optical Engineering; 19942005.
55. Centonze V, Pawley J. Tutorial on Practical Confocal Microscopy and use of the Confocal Test Specimen. In: Pawley JB,
editor. Handbook of Biological Confocal Microscopy. New
York: Plenum Press; 1995. p 549570.
56. Gratton E, vandeVen MJ. Laser Sources for Confocal
Microscopy. In: Pawley JB, editor. Handbook of Biological
Confocal Microscopy. New York: Plenum Press; 1995. p 6998.
57. Ashkin A, Dziedzic JM, Yamane T. Optical Trapping and
Manipulation of Single Cells using Infrared Laser Beams.
London 1987;330:769771.
58. DeMaggio S. Running and Setting Up a Confocal Microscope
Core Facility. In: Matsumoto B, editor. Cell Biological Applications of Confocal Microscopy. Methods in Cell Biology, Vol.
70. New York: Academic Press; 2002. p 475486.
59. Spring KR. Detectors for Fluorescence Microscopy. In: Periasamy A, editor. Methods in Cellular Imaging. New York:
Oxford University Press; 2001. p 4052.
60. Art J. Photon Detectors for Confocal Microscopy. In: Pawley
JB, editor. Handbook of Biological Confocal Microscopy. New
61. Amos WB. Instruments for Fluorescence Imaging. In: Allan
VJ, editor. Protein Localization by Fluorescence Microscopy: A
Practical Approach. New York: Oxford University Press; 2000.
pp 67108.
62. Hergert E. Detectors: Guideposts on the Road to Selection.
Photonics Design and Applications Handbook. 2001. pp
H110H113.
63. Piston DW, Patterson GH, Knobel SM. Quantitative Imaging
of the Green Fluorescent Protein (GFP). In: Sullivan KF, Kay
SA, editors. Green Fluorescent Proteins, Methods in Cell
Biology, Vol. 58. New York: Academic Press; 1999. pp 3147.
64. Carter DR. Photomultiplier Handbook: Theory, Design,
Application. Lancaster (PA): Burle Industries, Inc.; 1980.
65. Cody SH, et al. A Simple Method Allowing DIC Imaging in
Conjunction with Confocal Microscopy. J Microsc 2005;217:
265274.
66. Chang IC. Acousto-Optic Devices and Applications. In: Bass
M, Van Stryland EW, Williams DR, Wolfe WL, editors. Optics
II: Fundamentals, Techniques, and Design. New York:
McGraw-Hill; 1995. pp 12.112.54.
67. Wachman ES. Acousto-Optic Tunable Filters for Microscopy.
In: Yuste R, Lanni F, Konnerth A, editors. Imaging Neurons:
A Laboratory Manual. New York: Cold Spring Harbor Press;
2000. p 4.14.8.
68. Shonat RD, et al. Near-Simultaneous Hemoglobin Saturation
and Oxygen Tension Maps in Mouse Brain using an AOTF
Microscope. Biophy J 1997;73:12231231.
69. Wachman ES, Niu W, Farkas DL. Imaging Acousto-Optic
Tunable Filter with 0.35-Micrometer Spatial Resolution. App
Opt 1996;35:52205226.
70. Wachman ES. AOTF Microscope for Imaging with Increased
Speed and Spectral Versatility. Biophys J 1997;73:12151222.
476
71. Chen Y, Mills JD, Periasamy A. Protein Localization in

Living Cells and tissues using FRET and FLIM. Differentiation 2003;71:528541.
72. Wallrabe H, Periasamy A. Imaging Protein Molecules using
FRET and FLIM Microscopy. Curr Opin Biotech 2005;16: 1927.
73. Day RN, Periasamy A, Schaufele F. Fluorescence Resonance
Energy Transfer Microscopy of Localized Protein Interactions in the Living Cell Nucleus. Methods 2001;25:418.
74. Patterson GH, Lippincott-Schwartz J. A Photoactivatable
GFP for Selective Photolabeling of Proteins and Cells.
Science 2002;297:18731877.
75. Verkhusha VV, Lukyanov KA. The Molecular Properties and
Applications of Anthozoa Fluorescent Proteins and Chromoproteins. Nature Biotechnol 2004;22:289296.
76. Ando R, et al. An Optical Marker Based on the UV-Induced
Green-to-Red Photoconversion of a Fluorescent Protein. Proc
Natl Acad Sci USA 2002;99:1265112656.
77. Sharma D. The Use of an AOTF to Achieve High Quality
Simultaneous Multiple Label Imaging. Bio-Rad Technical
Notes. San Francisco: Bio-Rad, Note 4; 2001.
78. Miyawaki A, Sawano A, Kogure T. Lighting up Cells: Labeling
Proteins with Fluorophores. Nature Cell Biol 2003;5:S1S7.
79. Zhang J, Campbell RE, Ting AY, Tsien RY. Creating New
Fluorescent Probes for Cell Biology. Nature Rev Mol Cell Bio
2002;3:906918.
80. Lippincott-Schwartz J, Patterson G. Development and Use of
Fluorescent Protein Markers in Living Cells. Science
2003;300:8791.
81. Klonis N, et al. Fluorescence Photobleaching Analysis for the
Study of Cellular Dynamics. Eur Biophys J 2002;31:3651.
82. Lippincott-Schwartz J, Altan-Bonnet N, Patterson GH.
Photobleaching and Photoactivation: Following Protein
Dynamics in Living Cells. Nature Cell Biol 2003;5:S7S14.
83. Phair RD, Misteli T. Kinetic Modelling Approaches to in vivo
Imaging. Nature Rev Mol Cell Bio 2002;2:898907.
84. Politz JC. Use of Caged Fluorophores to Track Macromolecular
Movement in Living Cells. Trends Cell Biol 1999;9:284287.
85. Born M, Wolf E. Principles of Optics. New York: Cambridge
University Press; 1999.
86. Stelzer EHK. Contrast, Resolution, Pixelation, Dynamic range,
and Signal-to-Noise Ratio: Fundamental Limits to Resolution in
Fluorescence Light Microscopy. J Microsc 1997;189:1524.
87. Pawley J. Fundamental Limits in Confocal Microscopy. In:
Pawley JB, editor. Handbook of Biological Confocal Microscopy. New York: Plenum Press; 1995. p 1937.
88. Webb RH, Dorey CK. The Pixelated Image. In: Pawley JB,
editor. Handbook of Biological Confocal Microscopy. New
89. Jonkman JEN, Stelzer EHK. Resolution and Contrast in
Confocal and Two-Photon Microscopy. In: Diaspro A, editor.
Confocal and Two-Photon Microscopy: Foundations, Applications, and Advances. New York: Wiley-Liss; 2002. p 101125.
90. Haugland RP. The Handbook: A Guide to Fluorescent Probes
and Labeling Technologies. Chicago: Invitrogen Molecular
Probes; 2005.
91. Lemasters JJ, et al. Confocal Imaging of Ca2, pH, Electrical
Potential and Membrane Permeability in Single Living Cells.
92. Johnson I. Fluorescent Probes for Living Cells. Histochem J
1998;30:123140.
93. Kasten FH. Introduction to Fluorescent Probes: Properties,
History, and Applications. In: Mason WT, editor. Fluorescent
and Luminescent Probes for Biological Activity. New York:
Academic Press; 1999. p 1739.
94. Coons AH, Creech HJ, Jones RN, Berliner E. Demonstration
of Pneumococcal Antigen in Tissues by use of Fluorescent
Antibody. J Immunol 1942;45:159170.
95. Tsien RY. Building and Breeding Molecules to Spy on Cells

and Tumors. FEBS Lett 2005;579:927932.
96. Bruchez Jr M, et al. Semiconductor Nanocrystals as fluorescent Biological Labels. Science 1998;218:20132016.
97. Tsien RY, Waggoner A. Fluorophores for Confocal Microscopy. In: Pawley JB, editor. Handbook of Biological Confocal
Microscopy. New York: Plenum Press; 1995. p 267280.
98. Wessendorf MW, Brelje TC. Which Fluorophore is Brightest?
A Comparison of the Staining Obtained Using Fluorescein,
Tetramethylrhodamine, Lissamine Rhodamine, Texas Red
and Cyanine 3.18. Histochemistry 1992;98:8185.
99. Entwistle A, Noble M. The use of Lucifer Yellow, BODIPY,
FITC, TRITC, RITC and Texas Red for Dual Immunofluorescence Visualized with a Confocal Scanning Laser Microscope. J Microsc 1992;168:219238.
100. Darzynkiewicz Z. Differential Staining of DNA and RNA in
Intact Cells and Isolated Cell Nuclei with Acridine Orange.
Methods Cell Biol 1990;33:285298.
101. Waring MJ. Complex Formation Between Ethidium Bromide
and Nucleic Acids. J Mol Biol 1965;13:269282.
102. Arndt-Jovin DJ, Jovin TM. Fluorescence Labeling and Microscopy of DNA. Fluores Microsc Living Cells Culture Part B
103. Kubista M, Akerman B, Norden B. Characterization of Interaction between DNA and 40 ,6-Diamidino-2-phenylindole by
Optical Spectroscopy. Biochemistry 1987;26:45454553.
104. Loewe H, Urbanietz J. Basic Substituted 2,6-Bisbenzimidazole Derivatives: A Novel Series of Substances with Chemotherapeutic Activity. Arzneim-Forsch 1974;24:19271933.
105. Arndt-Jovin DJ, Jovin TM. Analysis and Sorting of Living
Cells According to Deoxyribonucleic Acid Content. J Histochem Cytochem 1977;25:585589.
106. Durand RE, Olive PL. Cytotoxicity, Mutagenicity and DNA
Damage by Hoechst 33342. J Histochem Cytochem 1982; 30:111116.
107. Panchuk-Voloshina N, et al. Alexa Dyes, A Series of New
Fluorescent Dyes that Yield Exceptionally bright, Photostable
Conjugates. J Histochem Cytochem 1999;47:11791188.
108. Berlier JE, et al. Quantitative Comparison of Long-Wavelength
Alexa Fluor Dyes to Cy Dyes: Fluorescence of the Dyes and
their Conjugates. J Histochem Cytochem 2003;51:16991712.
109. Mujumdar RB, et al. Cyanine Dye Labeling Reagents: Sulfoindocyanine Succinimidyl Esters. Bioconjugate Chem 1993;4:105111.
110. Ballou B, et al. Tumor Labeling in vivo using CyanineConjugated Monoclonal Antibodies. Cancer Immunol Immunother 1995;41:257263.
111. Zorov DB, Kobrinsky E, Juhaszova M, Sollott SJ. Examining
Intracellular Organelle Function Using Fluorescent Probes.
Circul Res 2004;95:239252.
112. Stephens DG, Pepperkok R. The Many Ways to Cross the Plasma
Membrane. Proc Natl Acad Sci USA 2001;98:42954298.
113. Rudolf R, Mongillo M, Rizzuto R, Pozzan T. Looking Forward to
Seeing Calcium. Nature Rev Mol Cell Biol 2003;4:579586.
114. Martin H, Bell MG, Ellis-Davies GC, Barsotti RJ. Activation
Kinetics of Skinned Cardiac Muscle by Laser Photolysis of
Nitrophenyl-EGTA. Biophys J 2004;86:978990.
115. White C, McGeown G. Imaging of Changes in Sarcoplasmic Reticulum [Ca2] using Oregon Green BAPTA 5N and Confocal Laser
Scanning Microscopy. Cell Calcium 2002;31:151159.
116. Helm PJ, Patwardhan A, Manders EM. A Study of the Precision of Confocal, Ratiometric, Fura-2-Based [Ca2] Measurements. Cell Calcium 1997;22:287298.
117. Rijkers GT, Justement LB, Griffioen AW, Cambier JC.
Improved Method for Measuring Intracellular Ca with
Fluo-3. Cytometry 1990;11:923927.
118. Schild D, Jung A, Schultens HA. Localization of Calcium Entry
through Calcium Channels in Olfactory Receptor Neurons
using a Laser Scanning Microscope and the Calcium Indicator
Dyes Fluo-3 and Fura-Red. Cell Calcium 1994;15: 341348.
119. Willoughby D, Thomas RC, Schwiening CJ. Comparison of
Simultaneous pH Measurements made with 8-Hydroxypyrene-1,3,6-trisulphonic acid (HPTS) and pH-Sensitive Microelectrodes in Snail Neurons. Pflugers Arch 1998;436:615622.
120. Ozkan P, Mutharasan R. A Rapid Method for Measuring
Intracellular pH using BCECF-AM. Biochim Biophys Acta
2002;1572:143.
121. Cody SH, et al. Intracellular pH Mapping with SNARF-1
and Confocal Microscopy. I: A Quantitative Technique for
Living Tissue and Isolated Cells. Micron 1993;24:573580.
122. Dubbin PN, Cody SH, Williams DA. Intracellular pH Mapping
with SNARF-1 and Confocal Microscopy. II: pH Gradients
within Single Cultured Cells. Micron 1993;24:581586.
123. Poot M, et al. Analysis of Mitochondrial Morphology and
Function with Novel Fixable Fluorescent Stains. J Histochem Cytochem 1996;44:13631372.
124. Keij JF, Bell-Prince C, Steinkamp JA. Staining of Mitochondrial
Membranes with 10-Nonyl Acridine Orange, MitoFluor Green,
and MitoTracker Green is Affected by Mitochondrial Membrane
Potential Altering Drugs. Cytometry 2000;39: 203210.
125. Reers M. et al. Mitochondrial Membrane Potential Monitored by JC-1 Dye. Methods Enzymol 1995;260:406417.
126. Price OT, Lau C, Zucker RM. Quantitative Fluorescence of 5FU-Treated Fetal Rat Limbs using Confocal Laser Scanning
Microscopy and LysoTracker Red. Cytometry 2003;53A:921.
127. Kumar RK, Chapple CC, Hunter N. Improved Double
Immunofluorescence for Confocal Laser Scanning Microscopy. J Histochem Cytochem 1999;47:12131217.
128. Suzuki T, Fujikura K, Higashiyama T, Takata K. DNA
Staining for Fluorescence and Laser Confocal Microscopy.
J Histochem Cytochem 1997;45:4953.
129. Haugland RP. Coupling of Monoclonal Antibodies with
Fluorophores. In: Davis WC, editor. Monoclonal Anibody
Protocols, Methods in Molecular Biology. Vol. 45. Totowa,
(NJ): Humana Press; 1995. p 205221.
130. Pagano RE, Martin OC. Use of Fluorescent Analogs of
Ceramide to Study the Golgi Apparatus of Animal Cells.
In: Celis JE, editor. Cell Biology: A Laboratory Handbook.
Vol. 2, 1998. p 507512.
131. Cole L, Davies D, Hyde GJ, Ashford AE. ER-Tracker Dye and
BODIPY-Brefeldin A Differentiate the Endoplasmic Reticulum
and Golgi Bodies from the Tubular-Vacuole System in Living
Hyphae of Pisolithus tinctorius. J Microsc 2000;197:239249.
132. Jaiswal JK, Mattoussi H, Mauro JM, Simon SM. Long-Term
Multiple Color Imaging of Live Cells using Quantum Dot
Bioconjugates. Nature Biotechnol 2003;21:4752.
133. Larson DR, et al. Water Soluble Quantum Dots for Multiphoton
Fluorescence Imaging in vivo. Science 2003;300: 14341436.
134. Watson A, Wu X, Bruchez M. Lighting up Cells with Quantum Dots. Biotechniques 2003;34:296303.
135. Michalet X, et al. Quantum Dots for Live Cells, in vivo
Imaging, and Diagnostics. Science 2005;307:538544.
136. Gao X, et al. In vivo Molecular and Cellular Imaging with
Quantum Dots. Curr Opin Biotech 2005;16:6372.
137. Lacoste TD, et al. Ultrahigh-Resolution Multicolor Colocalization of Single Fluorescent Probes. Proc Natl Acad Sci
USA 2000;97:94619466.
138. Tsien RY. The Green Fluorescent Protein. Ann Rev Biochem
1998;67:509544.
139. Sullivan KF, Kay SA, editors. Green Fluorescent Proteins, Methods in Cell Biology. Vol. 58. New York: Academic Press; 1999.
140. Conn PM, editor. Green Fluorescent Protein, Methods in
Enzymology. Vol. 302. New York: Academic Press; 1999.
477
141. Hicks BW, editor. Green Fluorescent Protein, Methods in

Molecular Biology. Vol. 183. Totowa (NJ): Humana Press; 2002.
142. Chalfie M, Kain S, editors. Green Fluorescent Protein: Properties, Applications, and Protocols. New York: Wiley-Liss; 1998.
143. Zimmer M. Green Fluorescent Protein: Applications, Structure, and Related Photophysical Behavior. Chem Rev
2002;102:759781.
144. Chalfie M, et al. Green Fluorescent Protein as a Marker for
Gene Expression. Science 1994;263:802805.
145. Heim R, Cubitt AB, Tsien RY. Improved Green Fluorescence. Nature (London) 1995;373:664665.
146. Heim R, Prasher DC, Tsien RY. Wavelength Mutations and
Posttranslational Autoxidation of Green Fluorescent Protein. Proc Natl Acad Sci USA 1994;91:1250112504.
147. Heim R, Tsien RY. Engineering Green Fluorescent Protein
for Improved Brightness, Longer Wavelengths, and Fluorescence Resonance Energy Transfer. Curr Biol 1996;6:
178182.
148. Wachter RM, et al. Structural Basis of Spectral Shifts in the
Yellow-Emission Variants of Green Fluorescent Protein.
Structure 1998;6:12671277.
149. Matz MV, et al. Fluorescent Proteins from Nonbioluminescent Anthozoa Species. Nature Biotechnol 1999;17:969973.
150. Matz MV, Lukyanov KA, Lukyanov SA. Family of the Green
Fluorescent Protein: Journey to the End of the Rainbow.
BioEssays 2002;24:953959.
151. Natural Animal Coloration can be Determined by a Nonfluorescent Green Fluorescent Protein Homolog. J Biol
Chem 2000;275:2587925882.
152. Song L, Hennink EJ, Young IT, Tanke HJ. Photobleaching
Kinetics of Fluorescein in Quantitative Fluorescence Microscopy. Biophys J 1995;68:25882600.
153. Berrios M, Conlon KA, Colflesh DE. Antifading Agents for
Confocal Fluorescence Microscopy. Methods Enzymol 1999;
307:5579.
154. Lakowicz JR. Principles of Fluorescence Spectroscopy. New
York: Kluwer Academic/Plenum Publishers; 1999.
155. Song L, Varma CA, Verhoeven JW, Tanke HJ. Influence of
the Triplet Excited State on the Photobleaching Kinetics of
Fluorescein in Microscopy. Biophys J 1996;70:29592968.
156. Herman B. Fluorescence Microscopy. New York: BIOS
Scientific Publishers; 1998.
157. Bunting JR. A Test of the Singlet Oxygen Mechanism of
Cationic Dye Photosensitization of Mitochondrial Damage.
Photochem Photobiol 1992;55:8187.
158. Byers GW, Gross S, Henrichs PM. Direct and Sensitized
Photooxidation of Cyanine Dyes. Photochem Photobiol 1976;
23:3743.
159. Dittrich PS, Schwille P. Photobleaching and Stabilization of
Fluorophores used for Single Molecule Analysis with Oneand Two-Photon Excitation. Appl Phys B 2001;73:829837.
160. Gandin E, Lion Y, Van de Vorst A. Quantum Yield of Singlet
Oxygen Production by Xanthene Derivatives. Photochem
Photobiol 1983;37:271278.
161. Kanofsky JR, Sima PD. Structural and Environmental
Requirements for Quenching of Singlet Oxygen by Cyanine
Dyes. Photochem Photobiol 2000;71:361368.
Further Reading
Willison JR. Signal Detection and Analysis, In: Bass, M. Van
Stryland, E. W. Williams DR, Wolfe WL. Optics I: Fundamentals,
Techniques, and Design. New York: McGraw-Hill; pp 1995. 18.1
18.16.
See also C ELLULAR IMAGING ; FLUORESCENCE
ION-SENSITIVE FIELD EFFECT TRANSISTORS.
MEASUREMENTS ;
478
MAHROKH DADSETAN
LICHUN LU
MICHAEL J. YASZEMSKI
Mayo Clinic,
College of Medicine
Rochester, Minnesota
INTRODUCTION
Invention of the light microscope by Janssens in 1590 was
the first milestone in the microscopic world. Janssens
microscope magnified objects up to 2030 times their original size. By the beginning of the twentieth century,
objects could be magnified only up to 1000 times with a
resolution of 0.2 mm. In the early 1930s, the limitations of
light microscopes and the scientific desire to see intracellular structural details, such as mitochondria and nuclei
led to the development of electron microscopes. The electron microscope took advantage of the much shorter
wavelength of the electron compared to that of visible light.
With the electron microscope, another 1000-fold increase
in magnification was accomplished with a concomitant
increase in resolution, allowing visualization of viruses,
deoxyribonuclic acid (DNA), and smaller objects, such as
molecules and atoms. The transmission electron microscope (TEM) was the first type of electron microscope,
and was developed by Ruska and Knoll in Germany in
1931. Electron microscopy is based on a fundamental
physics concept stated in the de Broglie theory (1924). This
concept is that moving electrons have the properties of
waves. The second major advancement in electron microscopy was made by Busch, who demonstrated in 1926 that
electrostatic or magnetic fields could be used as a lens to
focus an electron beam. In 1939, Siemens Corp. began
commercial production of a microscope developed by Von
Borries and Ruska in Germany. Hiller, Vance and others
constructed the first TEM in North America in 1941. This
instrument had a resolution of 2.5 nm.
About the same time that the first TEM was nearing
completion in the 1930s, a prototype of the scanning electron microscope (SEM) was constructed by Knoll and Von
Ardenne in Germany. However, the resolution of this
microscope was no better than that of the light microscope.
Following several improvements made by RCA in the
United States, as well as Cambridge University in England, a commercial SEM became available in 1963. A later
version of the SEM made by the Cambridge Instrument Co.
had a resolving power of 2050 nm and a useful magnification of 75,000. Recent models of the SEM have a
resolving power of 3.0 nm and magnifications up to
300,000.
Although the design of TEM and SEM is similar in many
ways, their applications are very different. The TEM is
patterned after as light microscope, except that electrons
instead of light pass through the object. The electrons are
then focused by two or more electron lenses to form a
greatly magnified image onto photographic film or a charge
coupled device (CCD) camera. The image produced by TEM
is two-dimensional (2D) and the brightness of a particular
region of the image is proportional to the number of electrons

that are transmitted through the specimen at that position
on the image. The SEM produces a three-dimensional (3D)
image by scanning the surface of a specimen with a 23 nm
spot of electrons to generate secondary electrons from the
specimen that are then detected by a sensor. The resolution
of an SEM is limited by two quite different sets of circumstances. One of these is concerned with the physics of
electron optics, while the other depends on the penetration
of electrons into the object being imaged.
A third type of electron microscope, the scanning transmission electron microscope (STEM) has features of both
the transmission and scanning electron microscopes. This
microscope is an analytical tool that determines the presence and distribution of the atomic elements in the
specimen. Recently, two groups of researchers have accomplished a subangstrom resolution (0.06 nm) for STEM
using an aberration corrector. They have reported that
columns of atoms in a silicone crystal that are 0.078 nm
apart can be distinguished at this resolution (1). The image
of Si shown in Fig. 1 has been recorded in a high angle
annular dark field (HAADF) mode, and the pairs of atomic
columns are seen directly resolved. The HAADF detector
collects electrons scattered by the sample to angles greater
than the detector inner radius. Such high angle scattering
is largely incoherent thermal diffuse scattering, which
means that the resolution observed in the image is determined by the intensity distribution of the illuminating
probe. With this advantage over conventional coherent high
resolution transmission electron microscopy (HRTEM),
HAADFSTEM has enabled imaging not only of individual
atomic columns in crystals, but single dopant atoms on their
surface and within their interior.
In 1982, another type of electron microscope, the scanning tunneling microscope (STM) was developed by two
scientists, Rohrer and Binnig, for studying surface structure. This invention was quickly followed by the development of a family of related techniques classified as
scanning probe microscopy (SPM). These techniques are
based upon moving a probe (typically called a tip in STM,
which is literally a sharp metallic object) just above a
specimens surface while monitoring some interaction
Figure 1. Image of a silicone crystal observed in the [112]

orientation recorded with an aberration corrected STEM. (From
Ref. 1). Reproduced by courtesy of American Association for the
Advancement of Science.)
between the probe and the surface. Atomic force microscopy (AFM) is another important technique of this kind
but is not categorized as electron microscopy. Bennig and
Rohrer were awarded one-half of the 1986 Nobel Prize in
physics for invention of the STM, while Ernst Ruska was
awarded the other one-half of that same Nobel Prize for his
1931 invention of the electron microscope. In STM, electrons are speeded up in a vacuum until their wavelength is
extremely short, only one hundred-thousandth that of
white light. Beams of these fast-moving electrons are
focused on a cell sample and are absorbed or scattered
by the cells parts so as to form an image on an electronsensitive photographic plate. The STM is widely used in
both industrial and academic research to obtain atomic
scale images of metal surfaces. It provides a 3D profile of
the surface, which is useful in characterizing surface
roughness and determining the size and conformation of
surface molecules. (2) Invention of electron microscopy had
an enormous impact in the field of biology, specifically in
cell and tissue analysis. Almost 15 years after the invention
of the first electron microscope by Ruska, many efforts were
made to apply this technique to biological problems. Using
the electron microscope, cell organelles and cell inclusions
were discovered or resolved in finer details. Electron microscopy, specifically TEM, is now among the most important
tools in cell biology and diagnostic pathology.
The latest advancement in electron microscopy is 3D
reconstruction of cellular components at a resolution that
is on the order of magnitude of atomic structures defined by
X-ray crystallography. The method for reconstruction of 3D
images of single, transparent objects recorded by TEM is
called electron tomography (ET). In order to generate 3D
images of individual molecules, one needs to obtain as
many tilt images as possible, covering the widest possible
angular range. The representative images of particles
obtained from different orientations is then analyzed
and combined by a software program to reconstruct the
molecule in 3D. With improvements in instrumentation,
data collection methods and techniques for computation,
ET may become a preferred method for imaging isolated
organelles and small cells. So far, the electron tomography
method covers the resolution range of 2.55.0 nm. Data
obtained via electron tomography furnish a rich source of
quantitative information about the structural composition
and organization of cellular components. It offers the
opportunity to obtain 3D information on structural cellular
arrangements with a significantly higher resolution than
that provided by any other method currently available
(e.g., confocal laser microscopy) (3).
479
Light
from
distant
source
Pinhole
aperture
Circular Biconcave
limiting
lens
aperture
Viewing
screen
Figure 2. Diffraction of light waves.
original waves. This bending or spreading phenomenon

is known as diffraction. Diffracted waves interfere with the
initial waves, and the result is an image of the edge of the
object. The edge appears to have a series of bands or fringes
called Fresnel fringes running parallel to the edge (Fig. 2).
Thus, if a strong beam of light illuminates a pinhole in a
screen and thus a pinhole serves as a point source and the
light passing through is focused by an apertured perfect
lens on a second screen, the image obtained is not a
pinpoint of light, but rather a bright central disk surrounded by a diffuse ring of light. Even if monochromatic
light was used to illuminate the point source and was to
pass through a perfect lens, the image will not be a sharp
one, but rather a diffuse disk composed of concentric rings.
This type of image is known as an Airy disk after Sir George
Airy, who first described this pattern during the nineteenth century (Fig. 3) (4). To determine resolving power
(RP), it is important to know the radius of the Airy disk.
The radius of the Airy disk as measured to the first dark
ring (r) is expressed by following equation:
r
0:612 l
nsin a
In Eq. 1, l wavelength of illumination; n refractive

index of the medium between the point source and the lens,
relative to free space; a half the angle of the cone of light
from the specimen plane accepted by the front lens of
objective
THEORY OF ELECTRON MICROSCOPY

According to electromagnetic theory, a light source initiates a vibrational motion that transmits energy in the
direction of propagation. The wave motion of light is analogous to that produced by a stone thrown into a pool of
water. When the waves generated from throwing a stone
strike an object that has an opening or aperture, another
series of waves is generated from the edge of the object. The
result is a new source of waves that emerges with the
Figure 3. Airy disks generated by viewing three pinholes in a

light microscope. Magnification of micrograph is 1000. (From
Ref. 4. Reproduced by courtesy of Jones and Bartlett Publishers.)
480
The above equation can be shown in another form as:

d
0:612 l
NA
where NA (numerical aperture) n sin a and represents

the light gathering power of the lens aperture.
From the above equation, RP is defined as the minimum
distance that two objects can be placed apart and still be
seen as separate entities. Consequently, the shorter the
distance, the better (or higher) is the RP of the system. For
example, consider a light microscope using ultraviolet (UV)
light, which lies beyond the lower end of the visible spectrum (400 nm). Further specifications of this system
include a glass slide with standard immersion oil (refractive index, n 1.5), and sina= 0.87 (sine of a 648 angle,
representing one-half of the 1288 acceptance angle of a
glass lens. The theoretical resolution that can be attained
by this system is 0.2 mm. In other words, two points in the
specimen that are not separated by at least this distance
will not be seen as two distinct points, but will be observed
as a single blurred image. Since the values of sin a and n
cannot be significantly increased beyond the stated values,
the RP can most effectively be improved by reducing wavelength.
Electron Beams And Resolution
The concept that moving electrons might be used as an
illumination source was suggested by a tenet of the de
Broglie theory, that moving electrons have wave properties. The wavelength of this particle-associated radiation is
given by following equation:
l
h
mn
where m is the mass of the particle, v the velocity of

particle, and h is Plancks constant (6.626 1034J1).
For an electron accelerated by a potential of 60,000 V
(60 kV), the wavelength of the electron beam would be
0.005 nm, which is 100,000 times shorter than that for
green light. By using Eq. 1, a TEM with perfect lenses
would therefore in theory be able to provide a resolution of
0.0025 nm. In practice, the actual resolution of a modern
high resolution transmission electron microscope is closer
to 0.2 nm. The reason we are not able to achieve the nearly
100-fold better resolution of 0.002 nm is due to extremely
narrow aperture angles (1000 times smaller than that of
the light microscope) needed by the electron microscope
lenses to overcome a major resolution limiting phenomenon called spherical aberration. In addition, diffraction,
chromatic aberration and astigmatism all contribute to
decreased resolution in TEM, and need to be corrected to
achieve higher resolution. (5)
objective lens of 100 followed by an eyepiece of 10. The

magnification of TEM would be 0.25mm/0.25 nm. This is a
106 magnification, and corresponds to a 1000-fold increase
in resolution compared to a light microscope. An objective
lens of 100 is followed by an intermediate lens of 25,
and the final image is projected by a projector lens of 100.
Further magnification for critical focusing is obtained by
viewing the image on the fluorescent screen with a long
working distance binocular microscope of 10. The final
image is photographed at 250,000. The processed negative
is then enlarged a further 4 in a photographic enlarger.
This result in a final prints (the electron micrograph) at the
desired magnification of 1066.
Electromagnetic Lenses
An electromagnetic lens is generated by a coil of wire with a
direct current (dc) that passes through the coil. This electromagnetic coil is called a solenoid. It forms an axially and
radially symmetric magnetic field that converges to a point.
A divergent cone of electrons enters from a point source,
and thus forms a real image on the lens axis. An advantage
of electromagnetic lenses is that the focal length can be
made infinitely variable by varying the coil current. Therefore, both magnification and image focus can be adjusted by
controlling the lens current (Fig. 4) (7).
Lens Aberrations
Electron lenses are affected by all the aberrations of optical
lenses, such as spherical aberration, chromatic aberration,
astigmatism, and distortion. Spherical aberration results
from the geometry of both glass and electromagnetic lenses
such that rays passing through the periphery of the lens
are refracted more than rays passing along the axis. Spherical aberration may be reduced by using an aperture to
eliminate some of the peripheral rays. Although this aperture is attractive for reducing spherical aberration, it
decreases the aperture angle and thereby prevents the
electron microscope from achieving the theoretical resolution predicted by Eq. 1.
Chromatic aberration results when electromagnetic
radiations of different energies converge at different focal
Source of
electrons
Axis
Electron
trajectory
Focal
point
Magnification
The maximum magnification of any microscope is simply
the ratio of the microscopes resolution to the resolution
of the unaided human eye. The resolution of the eye viewing
an object at 25 cm is generally taken to be 0.25 mm. Since the
resolution of a light microscope is 0.25 mm, maximum
useful light magnification is 1000, obtainable from an
Magnetic
lens field
Figure 4. Single electron passing through electromagnetic lens.
The electron is focused by the magnetic field to follow a trajectory
that will converge at a defined focal point after it emerges from the
lens.
planes. Chromatic aberration results in the enlargement of

a focal point with a consequential loss of resolution. It can
be corrected by using a monochromatic source of electromagnetic radiation. This entails stabilizing the accelerating voltage to generate the electrons with same levels of
energy, and having a good vacuum to minimize the energy
loss of the electrons during their passage through the
transmission specimen. This effect can also be reduced
by decreasing the aperture of the objective lens. (6)
Astigmatism is caused by radial asymmetry in a lens,
giving rise to a focal length in one plane that is different
from that in another plane. The fabrication and maintenance of a lens that has a perfectly symmetric lens field is
not feasible in practice. Thus, it is necessary to correct
astigmatism by applying a radial symmetry compensator
device called a stigmator. This consists of an adjustable
electric or magnetic field that can be applied across the lens
in any chosen direction, thus compensating for astigmatism. Image distortion, due to magnification changing
across the field from the value at the center, may be
positive (called barrel distortion) or negative (called pincushion distortion). These effects can be compensated by
operating two lenses in series, arranging for barrel distortion in one to be compensated by pincushion distortion in the
other. The lens system in modern electron microscopes is
designed to automatically counterbalance the various types
of distortions throughout a wide magnification range. (4)
DESIGN OF THE TRANSMISSION ELECTRECTRON
MICROSCOPE
Both the light and electron microscopes are similar so far
as the arrangement and function of their components are
concerned. Thus, both microscopes, when used for photographic purposes, can be conveniently divided into the
following component systems.
Illuminating System
This system serves to produce the required radiation and to
direct it onto the specimen. It consists of source and condenser lenses.
Source Lens. The source of electrons, or cathode, is a
hairpin of fine tungsten wire about 2 mm long, maintained
at 2500 K by 2 W of alternting current (ac) or dc power.
Electrons boil off the white-hot tungsten surface, and are
shaped into a conical beam by an electrode system called
the gun. Two further electrodes, the shield and the anode,
combine to form an electrostatic collimating lens and accelerator. A suitable accelerating voltage (20100 kV) is chosen for the specimen under examination, and is applied to
the cathode as a negative potential so that the anode may
remain at earth potential. The cathode and shield are
therefore carried on an insulator. The filament-shield voltage (cathode bias) is made variable to adjust the total
current drawn from the filament, which in turn varies the
brightness of the final image (4).
The energy of the electrons in the TEM determines the
relative degree of penetration of electrons into a specific
sample, or alternatively, influences the thickness of mate-
481
rial from which useful information may be obtained. Thus,

a high energy TEM (400 kV) not only provides the highest
resolution but also allows for the observation of relatively
thick samples (e.g., 0.2 mm) when compared with the
more conventional 100 kV or 200 kV instruments. Because
of the high spatial resolution obtained, TEMs are often
employed to determine the detailed crystallography of finegrained, or rare, materials (6,8).
Condenser Lens. The condenser lens regulates the
convergence (and thus the intensity) of the illuminating
beam on the specimen. The divergent electron beam emerging from the anode aperture can, in simple instruments,
be used to illuminate the specimen directly. However,
sufficient image brightness for high magnification is difficult to obtain. As in the light microscope, a condenser
system is almost invariably interposed between the gun
and specimen to concentrate the beam on the region of the
specimen under examination. A single condenser lens suffices for electronoptical work up to 50,000. However, for
high resolution work a double condenser system is always
used, which will concentrate the beam into an area as small
as 1m diameter.
Specimen Manipulation System
The pierced metal grid carrying the specimen proper is
clamped at its periphery to a suitable holder designed to
conduct heat rapidly away. The specimen temperature in a
TEM may rise to 200 8C. The holder, including its attached
specimen grid is introduced into the evacuated specimen
chamber through an airlock by means of an insertion tool.
This tool is then generally withdrawn after the holder has
been placed on the translation stage. The holder is then
free to move with the stage, which is driven from outside
the column through airlocks, by means of levers and micrometer screws. Two mutually perpendicular stage movements, each of about 1 mm, allow any part of the grid
area to be brought to the microscope axis and viewed at the
highest magnification. Most instruments provide a scan
magnification so that the whole grid area may be viewed at
100. Suitable specimen areas are then chosen, and
centered for study at higher magnifications.
Imaging System
This part of the microscope includes the objective, intermediate, and projector lenses. It is involved in the generation of the image and the magnification and projection of
the final image onto a viewing screen or camera system.
Electrons transmitted by the specimen enter the objective
lens. Those passing through the physical aperture are
imaged at 100 in the intermediate lens object plane,
1020 cm below the specimen. The position of this primary
image plane is controlled by the objective lens current
(focus control). A second image, which may be magnified
or diminished, is formed by the intermediate lens, the
current through which controls overall magnification
(magnification control). This secondary image, formed in
the objective plane of the projector lens, is then further
magnified, and the overall magnification is determined by
the position of the fluorescent screen or film.
482
Image Recording System

The final image is projected onto a viewing screen coated
with a phosphorescent zinc-activated cadmium sulfide
powder. This powder is attached to the screen with a
binder, such as cellulose nitrate. Most electron microscopes
provide for an inclination of the viewing screen so that the
image may be conveniently examined either with the unaided
eye or through a stereomicroscope (the binoculars). Although
the stereomicroscope image may appear to be rough due to
the 100 mm sized phosphorescent particles that make up the
screen, it is necessary to view a magnified image in order to
focus accurately. Some microscopes may provide a second,
smaller screen that is brought into position for focusing. In
this case, the main screen remains horizontal, except during
exposure of the film. All viewing screens will have areas
marked to indicate where to position the image so that it will
be properly situated on the film. Preevacuated films are
placed into an air lock (camera chamber) under the viewing
screen and the chamber evacuated to high vacuum. The
chamber is then opened to the column to permit exposure
of the film. In modern electron microscopes (Fig. 5), exposure
is controlled by an electrically operated shutter placed below
the projector lens. As one begins to raise the viewing screen,
the shutter blocks the beam until the screen is in the appropriate position for exposure. The shutter is then opened for
the proper interval, after which the beam is again blocked
until the screen is repositioned.
Figure 5. Image of a 300 kV TEM (FEI-Tecnai G2 Polara) for

cryoapplications at liquid nitrogen and liquid helium temperatures.
(Reproduced by courtesy of FEI Company.)
DESIGN OF THE SCANNING ELECTRON MICROSCOPE

The SEM is made up of two basic systems, and the specimen is at their boundary. The first system is the electron
optical column that provides the beam of illumination that
is directed to the specimen. The second system consists of
the electron collection, signal amplification, and image
display units, which converts the electrons emitted from
the specimen into a visible image of the specimen.
Electron Optical Column
The electron gun and electron lenses are present in the
electron optical column of the SEM in an analogous fashion
to their presence in the TEM.
1. Electron Gun: The electron source is most commonly
the hairpin tungsten filament located in a triode
electron gun. The electrons are emitted by the filament (also called the cathode), and accelerated by a
field produced by the anode. The anode is usually at a
positive potential on the order of 15 kV with respect
to the cathode. A third electrode, the shield, lies
between the anode and cathode and is negative with
respect to the cathode. After leaving the bias shield
and forming an initial focused spot of electrons of
50 mm in diameter, a series of two to three condenser lenses are used to successively demagnify this
spot sometimes down to 2 nm. These small spot
sizes are essential for the resolutions required at
high magnifications. A heated tungsten filament is
the conventional electron source for most SEMs;
other special sources are lanthanum hexaboride
(LaB6) and the field emission guns (FEG). Both of
these latter sources produce bright beams of small
diameter and have much longer lifetimes than
heated tungsten filaments. Schottky emission has
largely replaced earlier source technologies based
on either tungsten and LaB6 emission or cold-field
emission in todays focused electron beam equipment
including SEM, TEM, Auger systems, and semiconductor inspection tools. Schottky and cold-field emission are superior to thermionic sources in terms of
source size, brightness and lifetime. Both are up to
1000 times smaller and up to 100 times brighter than
thermionic emitters.
2. Electron Lenses: Most SEMs have three magnetic
lenses in their column: the first, second, and final
condenser lenses. The first condenser lens begins the
demagnification of the 50 mm focused spot of electrons formed in the region of the electron gun. As the
amount of current running through the first condenser lens is increased, the focal length of the lens
becomes progressively shorter and the focused spot
of electrons becomes smaller. In our earlier discussion of electron lenses, it was noted that focusing
takes place by varying the focal length. This is
accomplished by changing the intensity of the lens
coil current, which in turn alters the intensity of the
magnetic field that is generated by the lens. As the
lens current increases, the lens strength increases
and the focal length decreases. A short focal length

lens consequently causes such a wide divergence of
the electrons leaving the lens that many electrons are
not able to enter the next condenser lens. The overall
effect of increasing the strength of first condenser
lens is to decrease the spot size, but with a loss of
electrons. An aperture is positioned in the lenses to
decrease the spot size and reduce spherical aberration by excluding the more peripheral electrons. Each
of the condenser lenses behaves in a similar manner
and possesses apertures.
In designing the final condenser lens, several performance characteristics must be considered:
(a) Aberrations. Since the intermediate images of the
crossover produced by the condenser lenses have
significantly larger diameters than the final spot
size, the effect of aberrations on these lenses are
relatively small. It is thus the effects of spherical
and chromatic aberration as well as the astigmatism of the final condenser lens that are critical in
the design and performance of the objective lens
of SEM.
(b) Magnetic Field. As a result of electron bombardment, secondary electrons in the SEM are
emitted over a wide solid angle. These have energies of only few electron volts, yet they must be
able to reach the detector to produce the necessary signal. As a result, the magnetic field at the
specimen must be designed so that it will not
restrict effective secondary electron collection.
(c) Focal Length. The extent of lens aberrations is
dependent upon the focal length. Thus, it is desirable to keep the latter as short as possible in order
to help minimize the effects of aberrations.
The final lens usually has externally adjustable
apertures. Normally, final apertures on the order
of 5070 mm are used to generate smaller, less electron dense spots for secondary electron generation
and imaging. Larger apertures, for example, 200 mm,
are used to generate larger spots with greater numbers of electrons. These large spots contain a great
deal of energy and may damage fragile specimens.
They are used primarily to generate X rays for elemental analysis rather than for imaging purposes.
Specimen Manipulation System
The specimen is normally secured to a metal stub and is
grounded to prevent the build up of static high voltage
charges when the beam electrons strike the specimen.
In order to orient the specimen precisely, relative to the
electron beam and electron detectors, all SEMs have controls for rotating and traversing the specimen in x, y, and
z directions. It is also possible to tilt the specimen in order
to enhance the collection of electrons by a particular
detector. These movements have a large effect on magnification, contrast, resolution and depth of field. Some
improvement can be made in imaging by reorientation
of the specimen.
483
Interaction Of Electron Beam With Specimen

Three basic possibilities exist as to the nature of the beam
specimen interaction used to generate the image:
1. Some primary electrons, depending on the accelerating voltage, penetrate the solid to depths as much as
10 mm. The electrons scatter randomly throughout
the specimen until their energy is dissipated by
interaction with atoms of the specimen.
2. Some primary electrons collide with or pass close to
the nucleus of an atom of the specimen such that
there is a change in the electrons momentum. This
results in electron scatter through a large angle and
electron reflection from the specimen. Such elastically reflected primary electrons are known as backscattered electrons.
3. Some primary electrons interact with the host atoms
so that as a result of collisions, a cascade of secondary
electrons is formed along the penetration path. Secondary electrons have energy ranges of 050 eV and
are the electrons most commonly used to generate
the 3D image. The mean path length of secondary
electrons in many materials is 1 nm. Thus,
although electrons are generated throughout the
region excited by the incident beam, only those electrons that originate <1 nm deep in the sample escape
to be detected as secondary. The shallow depth of
production of detected secondary electrons makes
them very sensitive to topography.
In addition to producing backscattered and secondary
electrons, specimenbeam interactions also produce
photons, specimen currents, Auger electrons and X rays
that are characteristic of the probed specimen. These
emanations can be detected by X ray or electron spectroscopy for elemental analysis of the specimen surface. However, it is rarely used for biological specimens.
Signal Versus Noise
The signals generated as a result of the electron beam
striking a specimen are used to convey different types of
information about the specimen. In the usual SEM imaging
mode, signals consist of the secondary electrons generated
from the spot struck by the electron beam and noise consists of secondary electrons originating at locations away
from where the beam struck the specimen. The image
quality is eventually expressed by the signal-to-noise
(S/N) ratio. In a poor quality image the signal to noise
ratio is low. One may achieve a better image by either
reducing the noise or raising the signal. Since it is more
difficult to reduce the noise level, the signal is usually
raised by increasing the electron emissions from the
gun. Several methods to accomplish increased electron
emissions include: altering the bias settings, decreasing
the distance between the anode and the filament, decreasing the distance between the filament and the shield
aperture, and using either a lanthanum hexaboride filament or a cold-field emissions gun. A second method to
increase the signal is to use slower scan rates on the
484
specimen. Longer dwell times of the beam on the specimen

will generate more secondary electrons from the spot where
the beam strikes the specimen. This increase in current,
however, carries with it an increased risk of damage to
sensitive specimens (9).
Secondary Electron Detection
To collect the secondary electrons, a suitable electrode is
held at a positive potential and serves to attract them and
produce an emission current. The strength of this signal is
proportional to the number of electrons striking the collector. This signal is used, after amplification; to modulate
the intensity of the cathode-ray tube (CRT) beam as it
moves across the tube face, synchronously with the path of
the electron probe across the specimen surface.
Typically, the secondary electron collector is based on
the original 1960 scintillator-photomultiplier design of
Everhart and Thoronley. In this system, the secondary
electrons are accelerated towards the scintillator by a
potential difference of a few hundred to a few thousands
volts. Upon hitting the scintillator, each electron produces
many photons that are guided by the light pipe to the
photomultiplier. Each photoelectron triggers a release of
two or more secondary electrons at the first electrode
(dynode) and process cascades, yielding from 100,000 to
50 million additional electrons. Thus, the photomultiplier
reconverts the light to an electron current and provides a
high degree of amplification that can be controlled by
variation of the voltage applied to the dynodes (8).
Image Recording System
The final magnified image in the SEM is formed on a CRT
or monitor. Unlike TEM, in which the electrons interact
directly with the photographic medium, SEM images are
most often photographed directly from the monitor through
the lens of either a 35 mm roll film camera or a larger
4 in.5 in. (10.6 cm 12.7 cm) sheet film camera. The camera shutter remains open as the electron beam slowly scans
across the specimen. A valuable addition to most SEMs is
the automatic data display that permits the generation of
informational data on the viewing and recording monitors.
With this accessory, experiment numbers, dates, accelerating voltages and magnifications may be displayed.
DIAGOSTIC ELECTRON MICROSCOPY
Electron microscopy excels as a diagnostic tool with
respect to the detection and identification of both abnormal tissue anatomy and the pathogens responsible for the
disease. The value of electron microscopy in difficult diagnostic situations has been demonstrated repeatedly, particularly when there is close coordination between the
pathologist and the attending clinician. Ultrastructural
study may be applied to a variety of substances including
biological materials. By examination of specially prepared
tissue sections, changes not perceived by light microscopy
can be identified, leading to improved diagnostic interpretations. For example, in certain kidney diseases, such as
nephrotic syndrome and Nil disease, the correct diagnosis
can be made only by these means, and this in turn affects the
selection of therapy. Similarly, certain neoplasms can be

identified definitively only through ultrstructural studies,
with obvious implications for treatment and prognosis.
Another area of growing importance is the identification of
viral particles in biological material. In some instances,
ultrastructural study is the only way to establish the presence of a viral infection, and in other instances a diagnosis
may be made earlier than by serological methods. The costs
of diagnostic electron microscopy are relatively small in light
of the benefits to patient care. The following are examples
highlighting the use of electron microscopy in the diagnosis
of certain diseases.
Neoplasms
Many neoplasms appear undifferentiated by light microscopy, but most show differentiation along one cell line or
another at the ultrastructural level. However, it is noteworthy that not all of the ultrastructureal criteria for identifying the cell type may be present in every neoplasm. As
expected, the more differentiated the neoplasm, the more
likely will its cells contain a broad complement of diagnostic
morphologic features. Usually, the ultrastructural findings
do allow the pathologist to make a definitive diagnosis when
interpreted in conjunction with the light microscopic picture, and in some cases with the histochemical and immunohistochemical results. The example that follows will
highlight the use of diagnostic electron microscopy in the
identification of carcinomas. Various types of carcinomas
have a number of distinguishing features, but one common
characteristic of all carcinomas is the presence of intracellular junctions, usually desmosomes and/or intermediate
junctions. The presence of lumens, microvilli, tight junctions, junctional complexes, basal lamina, secretory granules, prominent Golgi apparatus, and moderately
prominent rough endoplasmic reticulum are all suggestive
of adenocarcinoma in the differential diagnosis of a neoplasm (10). Figure 6 shows a pancreatic carcinoma, which
may arise from acinar cells, centrocinar cells, intercalated
duct cells, interlobular duct cells, interlobular duct cells and
main pancreatic duct cells. Adenocarcinomas arising from
main and interlobular ducts (mucinous cystadenocarcinomas) have cells similar to those of bile ducts and intestinal
epithelium; that is, the cytoplasm contains mucin granules,
and the free surface has microvilli filled with thin filaments
that anchor into the subjacent cytoplasm (1113).
Infectious Diseases
Bacteria. Diagnostic criteria for bacterial rods and cocci
are (1) the presence of an outer-cell wall, (2) the presence
flagella or pili (fimbria) on the outer surface of the cell, (3)
the presence of an inner-cell membrane, (4) a central
nuclear region (nucleoid), without a limiting membrane,
(5) dense cytoplasm composed mostly of ribosomes, and (6)
a varying number of vesicles formed from the inner-cell
membrane (mesosomes), storage vacuoles and endospores
(14). Figure 7 shows the bacteria in Whipple disease. The
rods are present both free and within macrophages.
Viruses. The distinct morphology of members of different viral families usually allows an agent to be assigned to
Figure 6. Ductal, mucinous cystadenocarcinoma (pancreas). In

this field, the neoplastic cells form a cystic lumen (L) lined by
innumerable microvilli. An intracytoplasmic lumen (IL), without
junctional complexes, is present in one cell. Some of the cells
lining the lumen have a rich collection of mucinous granules (M)
in their apical cytoplasm. Lateral cell borders show a switch-backing
pattern of interdigitation (arrows). (6800) (From Ref. 10.
Reproduced by courtesy of Springer-Verlag GmbH.)
a particular family. This morpho-diagnosis, combined with

clinical information is often sufficient to permit a provisional diagnosis and to initiate treatment and containment
protocols while waiting for other test results. Diagnostic
criteria that apply to viruses in general are (1) the presence
of intracellular and/or extracellular elliptical, stand-like,
round or polygonal structures measuring 20300 nm in
diameter; and (2) the identification of viral morphology,
consisting of a central, electron dense core (DNA-containing
nucleoid) and an outer shell (capsid), which may have more
than one layer (Fig. 8) (15).
Fungi. The electron microscopic diagnostic criteria for
fungi include the identification of mononucleated oval
yeast forms measuring 24 mm in diameter, with a thin
cell wall and no true capsule. These can be located either
extracellularly or intracellularly (16). A representative
fungus, Histoplasma capsulatum, is shown in Fig. 9. The
organisms have a clear halo between their visible cytoplasm and their thin cell wall.
Skeletal Muscle Diseases
The skeletal muscle responses to injury that are visible
with the electron microscope can be categorized as follows:
(1) alterations in the sarcolemma (e.g., discontinuities of
485
Figure 7. Whippls disease. The lamina propria of the jejunal

mucosa contained numerous Whipples type macrophages [M
macrophage nucleus) and bacteria rods (B)]. Most of the intact
bacilli are extracellular, whereas those in the macrophage are
in various stages of degeneration, including the end-stage of
serpiginous membrane (). (16,500 ) (From Ref. 10. Reproduced
by courtesy of Springer-Verlag GmbH.)
the plasma membrane or the basement membrane); (2)

alterations in myofilaments; (3) Z-band alterations (e.g.,
streaming and nemaline bodies); (4) nuclear changes (e.g.,
abnormal location of the nucleus within the muscle fiber
and nuclear inclusions); (5) abnormalities of the sarcoplasmic reticulum and the T-system (e.g., tubular aggregates),
(6) abnormal accumulations of metabolites (e.g., glycogen
and lipids); (7) abnormal cytoplasmic structures (e.g.,
vacuoles, cytoplasmic bodies, concentric laminated bodies,
fingerprint bodies, curvilinear bodies). In general, many of
these ultrastructural abnormalities are not specific for a
single disease. Electron microscopy can be a valuable
adjunct to help the pathologist arrive at the proper interpretation of a muscle biopsy when taken together with all
other available clinical, electrophysiologic, and histopathological data. In addition to the pathologic changes that
might involve the muscle fibers themselves, many diseases
of muscle also simultaneously affect adjoining connective
tissue components, blood vessels and intramuscular
nerves. It is therefore important to pay particular attention
to these structures when examining muscle with the light
and electron microscope (10). The light micrograph shown
in Fig. 10 demonstrates centrally placed nuclei in the
majority of the muscle fibers. The central nuclei often
are surrounded by a clear area that is devoid of adenosine
triphosphatase (ATPase) activity. Ultrastructural features
of the paranuclear clear zone in Fig. 11 include (1) the
486
Peripheral Nerve Diseases

Wallerian Degeneration.
Ultrastructural changes
detected in peripheral nerve specimens include either the
general pathologic responses of the peripheral nerve to
either the injury or the specific disease entity that afflicts
the patient. The general pathologic processes involving
peripheral nerve can be divided into two broad categories:
those that indicate a process primarily affecting the axon
and those that indicate a process primarily affecting the
myelin sheath. Examination of peripheral nerve biopsies by
electron microscopy therefore must include evaluation of the
axons, the interstitium, and the myelin and Schwann cells.
The sequence of structural changes following nerve
injury that are collectively called Wallerian degeneration
is shown in Fig. 12. Wallerian degeneration specifically
refers to degeneration of the distal segments of a peripheral
nerve after severance of the axons from their cell bodies
(17). When the nerve injury is a contusion, the basement
membrane of the Schwann cell is preserved, allowing
regeneration within the endoneurial tube. In contrast,
when the nerve injury is a transection, the endoneurial
tube (composed of denervated Schawnn cells and extracellular matrix) may not be appropriately aligned with the
regenerating axons. Axonal regeneration is therefore less
efficient after nerve degeneration that follows a transection
injury compared to that following a crush injury (10,17).
Figure 8. Herpes simplex encephalitis (cerebrum). Virions (V)
have a central dense nucleoid and an outer three-layered capsid.
(63,800) (From Ref. 10. Reproduced by courtesy of SpringerVerlag GmbH.)
absence of myofilaments, (2) numerous mitochondria, (3)

glycogen accumulation. In some patients, especially
infants, the central clear zone may be more evident than
the nuclei within that zone when examining the tissue in
cross-section (10).
Figure 9. Histoplasma capsulatum (supraclavicular lymph node). High magnification

of parasitic yeast forms illustrates details
of their internal structure. N nucleus;
* clear, peripheral, cytoplasmic halo;
arrows parasitic cell membrane. (20,000)
(From Ref. 10. Reproduced by courtesy of
Springer-Verlag GmbH.)
PROSPECTS OF ELECTRON MICROSCOPY

In the 1980s, electron microscopy lost much of its former
role in the life sciences due to the introduction of modern,
highly effective molecular analytical techniques, such as
immunohistochemistry, chip technology, and confocal
laser scan microscopy. In recent years, however, substantial technical improvements were made in specimen preparation, instrumentation and software, allowing the
487
Figure 10. Centronuclear (myotubular)

myopathy. The majority of small fibers
contain centrally placed nuclei. (H&E,
150) (From Ref. 10. Reproduced by
courtesy of Springer-Verlag GmbH.)
electron microscope to reemerge as a valuable tool for

analyzing molecular complexes.
Electron microscopy as an imaging technique allows a
direct view of biological objects, while some of the other
available techniques are indirect and in some instances
nonspecific. Using electron microscopy, all components of
the object and their mutual relationships at the molecular
level can be analyzed. This information provides insight
toward an understanding of structurefunction relations.
The possibility of 3D reconstruction of cellular components via electron microscopy, along with the ease and
speed with which newer instruments can provide data,
have given the way to what many in the field are referring as a revolution. Recent technological advancements
have made automated data acquisition possible, and have

thus allowed a reduction of the total electron dose needed
to image a specimen. Specimen preparation advances,
such as embedding biological specimens in vitreous ice,
have enabled studies of the macromolecular organization
of cells. Whole prokaryotic and small eukaryotic cells can
be directly grown and hydrated frozen on electron microscopy grids. Examination of the naturally preserved
cells delivers images of the cellular structures in their
functional environment. Such so-called tomograms contain all available information about the spatial relationships of macromolecular structures within the cell.
However, due to their poor S/N and the generally highly
crowded nature of the cytoplasm, the interpretation of
Figure 11. Muscle fiber with degenerated

nucleus. Note absence of myofilaments
and accumulation of glycogen in the paranuclear region to the right (15,000). (From
Ref. 10. Reproduced by courtesy of SpringerVerlag GmbH.)
488
BIBLIOGRAPHY
Figure 12. Ultrathin sections of opossums optic nerve fibers 24 h

after crash. Normal fibers (n) are seen among some altered fibers,
which exhibit watery degeneration (star) and myelin sheath
breakdown (thick arrow). Note demyelinated fibers (thin arrows)
with an apparently intact axoplasmic cytoskeleton. Asterisk, astrocytic processes. (From Ref. 17. Reproduced by courtesy of Anais da
Academia Brasileira de Ciencias.)
these tomograms remains difficult. To get significant

information about specific structures in the cell, the
images have to be evaluated using advanced pattern
recognition methods. Existing structural models of cellular constituents at lower resolutions can guide the systematic evaluation of the tomograms. The aim is to
visualize the complete 3D organization of the cell at
molecular resolution. Structural evaluation by single particle analysis, electron crystallography and electron tomography is slow compared to other structure determination
technologies, in particular X-ray crystallography. Processing time for the electronic technique is typically in the
range of several months per solved structure, depending
on the resolution achieved. The same task can be accomplished in the range of hours or days for X-ray crystallography, once suitable crystals are available. Continued
joint efforts between the research community and manufacturers to develop user-friendly, universal interfaces
between electron crystallography, single-particle analysis
and electron tomography would improve this situation, and
further expand the usefulness of of these electronic technologies.
1. Nellist PD, et al Direct sub-angstrom imaging of a crystal

lattice. Science 2004;305(5691):1741.
2. Freeman MR. Time -resolved scanning tunneling microscopy
through tunnel distance modulation. Appl Phys Lett
1993;68(19):26332635.
3. Bonetta L. Zooming in on electron tomography. Nature Methods 2005;2(2):13944.
4. Bozzola JJ, Russell LD. Electron Microscopy: Principals and
techniques for biologists. Sudbury (MA): Jones and Bartlett
Publishers; 1998.
5. Hayat MA. Principels and techniques of electron microscopy:
Biological application. New York: Van Nostrand Reinhold
Company; 1973.
6. Wischnitzer S. Introduction to electron microscopy. New
York: Pergamon Press; 1981.
7. Meek GA. Practical electron microscopy for biologists. New
York: John Wiley & Sons inc; 1976.
8. Joy DC. Beam interactions, contrast and resolution in the
SEM. J Microsc 1984;136:24158.
9. Haine M. The electron microscope: The present state of the
Art. London: Spon; 1961.
10. Dickersin GR. Diagnostic Electron Microscopy: A text/ atlas.
New York: Springer-Verlag; 1999.
11. Franchina M, Del Borrello E, Caruso A, Altavilla G. Serous
tumors of the ovary: Ultrastructural observations. Eur J
Gynaecol Oncol 1992;13(3):26876.
12. Wolf HK, Garcia JA, Bossen EH. Oncocytic differentiation in
intrahepatic biliary cystadenocarcinoma. Modern Pathol
1992;5(6):665866.
13. Kobayashi TK, et al. Effects of Taxol on ascites cytology from a
patient with fallopian tube carcinoma: Report of a case with
ultrastructural studies. Diagn Cytopathol 2002;27(2):132134.
14. Yogi T, et al Whipples disease: The first Japanese case
diagnosed by electron microscopy and polymerase chain reaction. Intern Med 2004;43(7):566570.
15. Jensen HL, Norrild B. Herpes simplex virus-cell interactions
studied by immunogold cryosection electron microscopy.
Methods Mol Biol 2005;292:143160.
16. Garrison RG, Boyd KS. Electron microscopy of yeastlike cell
development from the microconidium of Histoplasma capsulatum. J Bacteriol 1978;133(1):345353.
17. Narciso MS, Hokoc JN, Martinez AM. Watery and dark axons
in Wallerian degeneration of the opossums optic nerve:
Different patterns of cytoskeletal breakdown? An Acad Bras
Cienc 2001;73(2):231243.
See also ANALYTICAL METHODS, AUTOMATED; CELLULAR IMAGING; CYTOLOGY,
AUTOMATED.
SERGE PELET
MICHAEL PREVITE
PETER T. C. SO
Massachusetts Institute of
Technology
Cambridge, Massachusetts
INTRODUCTION
Fluorescence microscopy quantifies the distribution of
fluorophores and their biochemical environment on the
micron length scale and allows In vivo measurement of

biological structures and functions (13). Heimsta dt developed one of the earliest fluorescence microscopes in 1911.
Some of the first biochemical applications of this technique
include the study of living cells by the protozoologist Provazek in 1914.
Fluorescence microscopy is one of the most ubiquitous
tools in biomedical laboratories. Fluorescence microscopy
has three unique strengths. First, the fluorescence microscope has high biological specificity. Based on endogenous
fluorophores or exogenous probes, fluorescence microscopy
allows the association of a fluorescence signal with a specimen structural and biochemical state. While fluorescence
microscopy has comparable resolution to white light microscopes, their range of applications in biomedicine is much
broader.
Second fluorescence microscopy is highly sensitive in
the imaging of cells and tissues. The high sensitivity of
fluorescence microscopy originates from two factors. One
factor is the significant separation between the fluorophores excitation and emission spectra. This separation
allows the fluorescence signal to be detected by efficiently
rejecting the excitation radiation background using bandpass filters. The fluorescence microscope has the sensitivity
to image even a single fluorophore. The other factor is the
weak endogenous fluorescence background in typical biological systems. Since there is minimal background fluorescence, weak fluorescence signal from even a few
fluorescent exogenous labels can be readily observed.
Third, fluorescence microscopy is a minimally invasive
imaging technique. In vivo labeling and imaging procedures are well developed. While photodamage may still
result from prolonged exposure of shorter excitation radiation, long-term observation of biological processes is possible. Today, a single neuron in the brain of a small animal
can be imaged repeatedly over a period of months with no
notable damage.
SPECTROSCOPIC PRINCIPLES OF FLUORESCENCE
MICROSCOPY
489
vibrational level in an electronic excited state. The molecule is quickly relaxed to the lowest vibrational level of the
excited electronic state after excitation on the time scale of
femtoseconds to picoseconds via vibrational processes. The
energy loss in the vibrational relaxation process is the origin
of the Stoke shift where fluorescence photons have longer
wavelengths than the excitation radiation. The coupling of
the ground and excited state both for the absorption and
emission process is governed by the FranckCondon principle, which states that the probability of transition is
proportional to the overlap of the initial and final vibrational wave function. Since the vibrational level structures
of the excited and ground states are similar, the fluorescence emission spectrum is a mirror image of the absorption spectrum, but shifted to lower wavelengths. The shift
between the maxima of the absorption and emission spectra is referred to as the Stokes shift. The residence time of a
fluorophore in the excited electronic state before returning
to the ground state is called the fluorescence lifetime. The
fluorescence lifetime is typically on the order of nanoseconds. The Jablonski diagram represents fluorescence excitation and deexcitation processes (Fig. 1).
Fluorescence deexcitation processes can occur via radiative and nonradiative pathways. Radiative decay describes
molecular deexcitation processes accompanied by photon
emission. Molecules in the excited electronic states can also
relax by nonradiative processes where excitation energy is
not converted into photons, but are dissipated by thermal
processes, such as vibrational relaxation and collisional
quenching. Let G and k be the radiative and nonradiative
decay rates, respectively, and N be the number of fluorophore in the excited state. The temporal evolution of the
excited state can be described by
dN
G kN
dt
(1)
N N0 eGkt N0 et=t
(2)
S1
Fluorescence Spectroscopy
An understanding of spectroscopic principles is essential to
master fluorescence microscopy (46). Fluorescence is a
photon emission process that occurs during molecular
relaxation from electronic excited states. Historically,
Brewster first witnessed the phenomenon of fluorescence
in 1838 and Stokes coined the term fluorescence in 1852.
These photonic processes involve transitions between electronic and vibrational states of polyatomic fluorescent
molecules (fluorophores) by the absorption of either one
or more photons. Electronic states are typically separated
by energies on the order of 10,000 cm1 and vibrational
sublevels are separated by 102103 cm1. In a one-photon
excitation process, photons with energies in the ultraviolet
(UV) to the bluegreen region of the spectrum are needed to
trigger an electronic transition, whereas photons in the
infrared (IR) spectral range are required for two-photon
excitation. The molecules from the lowest vibrational level
of the electronic ground state are excited to an accessible
IC
T1
1p
2p
IS
F
S0
VL
Figure 1. A Jablonski diagram describing fluorescence (F) and

phosphorescence (P) emission and excitation processes based on
one-photon (1p) and two-photon (2p) absorption. The parameters
S0, S1, and T1 are the electronic singlet ground state, singlet
excited state, and triplet excited state, respectively. Here VL
denotes vibrational levels, IC denotes internal conversion, and
IS denotes intersystem crossing.
490
The fluorescence lifetime, t, of the fluorophore is

the combined rate of the radiative and nonradiative
pathways:
t
1
Gk
(3)
One can define the intrinsic lifetime of the fluorophore in

the absence of nonradiative decay processes as, t0:
t0
1
G
(4)
The efficiency of the fluorophore can then be quantified by

the fluorescence quantum yield, Q, which measures the
fraction of excited fluorophore relaxing via the radiative
pathway:
G
t
Q
(5)
G k t0
Environmental Effect on Fluorescence
A number of factors contributes to the nonradiative decay
pathways of the fluorophores and reduces fluorescence
intensity. In general, the nonradiative decay processes
can be classified as
k kic kec ket kis
(7)
where [Q] is the concentration of the quencher and k0 is

related to the diffusivity and the hydrodynamics radii of
the reactants.
When collisional quenching is the dominant nonradiative process, equation 1 predicts that fluorescence
lifetime decreases with quencher concentration.
t0
1 k0 t 0 Q
t
(8)
Collision quenching also reduces the steady-state fluorescence intensity, F, relative to the fluorescence intensity in
the absence of quencher, F0. The SternVolmer equation
describes this effect:
F0
1 k0 t 0 Q
F
F0
1 Ks Q
F
(9)
A related process is steady-state quenching, where fluorescence signal reduction is due to ground-state processes. A
(10)
where Ks is the association constant of the quencher and

the fluorophore. However, since steady-state quenching is
a ground-state process that only reduces the fraction of
fluorophores available for excitation, fluorescence lifetime
is not affected.
Resonance energy-transfer rate, ket, becomes significant
when two fluorophores are in close proximity within 5
10 nm as during molecular binding. The energy of an
excited donor can be transferred to the accepted molecule
via an induced dipoleinduced dipole interaction. Let D
represents the donor and A, the acceptor. Under illumination at the donor excitation wavelength, the number of
excited donors and acceptors are ND, NA, respectively.
Further, define the donor and acceptor deexcitation rates
as kD and kA. The excited-state population dynamics of the
donor and acceptor can be described as
(6)
where kic is the rate of internal conversion, kec is the rate

of external conversion, ket is the rate of energy transfer,
and kis is the rate of intersystem crossing.
Internal conversion describes the process where the
electronic energy is converted to thermal energy via a
vibrational process. The more interesting process is external conversion, where fluorophores lose electronic energy
in collision process with other solutes. Several important
solute molecules, such as oxygen, are efficient fluorescence
quenchers. The external conversion process provides a
convenient mean to measure the concentration of these
molecules in the microenvironment of the fluorophore. The
fluorophore is deexcited nonradiatively upon collision. The
collisional quenching rate can be expressed as
kec k0 Q
fluorophore can be chemically bound to a quencher to form

a dark complex, a product that does not fluoresce. In this
case, steady-state fluorescence intensity also decreases
with quencher concentration as
dN D
kD ket N D
dt
(11)
dN A
kA N A ket N D
dt
(12)
Solving these equations provides the dynamics of donor

and acceptor fluorescence:
N D N0D expkD ket t
N A N0D
(13)
ket
expkD t ket t expkA t (14)
kA kD ket
The donor decay is a shortened single exponential, but the

acceptor dynamics is more complex with two competing
exponential processes.
The intersystem crossing rate, kis, describes transitions
between electronic excited states with wave functions of
different symmetries. The normal ground state is a singlet
state with an antisymmetric wave function. Excitation of
the ground-state molecule via photon absorption results in
the promotion of the molecule to an excited state with an
antisymmetric wavefunction, another singlet state. Due to
spinorbit coupling, the excited molecule can transit into a
triplet sate via intersystem crossing. The subsequent
photon emission from the triplet state is called phosphorescence. Since the decay of the triplet state to the singlet
ground state is forbidden radiatively, the triplet excited
state has a very long lifetime on the order of microseconds
to milliseconds.
FLUORESCENCE MICROSCOPE DESIGNS
The components common to most fluorescence microscopes
are the light sources, the optical components, and the
detection electronics. These components can be configured
to create microscope designs with unique capabilities.
491
Fluorescence Excitation Light Sources

Fluorescence excitation light sources need to produce
photons with sufficient energy and flux level. The ability
to collimate the emitted rays from a light source further
determines its applicability in high resolution imaging.
Other less critical factors, such as wavelength selectivity,
ease of use, and cost of operation, should also be considered.
Mercury arc lamps are one of the most commonly used
light sources in fluorescence microscopy. The operation of a
mercury arc lamp is based on the photoemission from
mercury gas under electric discharge. The photoemission
from a mercury arc consists of a broad background punctuated by strong emission lines. A mercury lamp can be
considered as a quasimonochromatic light source by utilizing one of these strong emission lines. Since mercury lamps
have emission lines throughout the near-UV and visible
spectrum, the use of a mercury lamp allows easy matching
of the excitation light spectrum with a given fluorophore by
using an appropriate bandpass filter. Mercury arc lamps
are also low cost and easy to use. However, since the
emission of mercury lamps are difficult to collimate, they
are rarely used in high resolution techniques, such as
confocal microscopy. The advent of high power, energy
efficient, light-emitting diodes (LEDs) with a long operation life allows the design of new light sources that are
replacing arc lamps in some microscopy applications.
Laser light sources are commonly used in high resolution fluorescence microscopes. Laser light sources have a
number of advantages including monochromatcity, high
radiance, and low divergence. Due to basic laser physics,
the laser emission is almost completely monochromatic.
For fluorescence excitation, a monochromatic light source
allows very easy separation of the excitation light from the
emission signal. While the total energy emission from an
arc lamp may be higher than some lasers, the energy
within the excitation band is typically a small fraction of
the total energy. In contrast, lasers have high radiance: the
energy of a laser is focused within a single narrow spectral
band. Therefore, the laser emission can be more efficiently
used to trigger fluorescence excitation. Furthermore, laser
emission has very low divergence and can be readily collimated to form a tight focus at the specimen permitting
high resolution imaging. Gas lasers, such as the argonion
laser and heliumneon lasers, are commonly used in fluorescence microscopy. Nowadays, they tend to be replaced by
solid-state diode lasers that are more robust and fluctuate
less. Lasers can further be characterized as continuous
wave and pulsed. While continuous wave lasers are sufficient for most applications, pulsed lasers are used in twophoton microscopes where high intensity radiation is
required for efficient induction of nonlinear optical effects.
(3)
(1)
AP
(4)
(2)
Figure 2. Four basic rules of optical ray tracing. (1) Light

emerging from the focal point will become collimated parallel to
the optical axis after the lens. And inversely (2) a collimated beam
parallel to the optical axis will be focused at the focal plane of the
lens. (3) A light source in the focal plane of the lens will become
collimated after passing through the lens with an oblique angle
determined by the distance from the optical axis and inversely (4)
An oblique collimated beam will be focused in the focal plane by the
lens. The numerical aperture of an imaging system is a function of
the maximum convergence angle, a, as defined in rule 2. The
maximum convergence angle is a function of the lens property
and its aperture (AP) size.
originated from the focal plane of a lens will emerge

collimated. (4) Collimated light rays incident upon a lens
will focus at its focal plane (Fig. 2). From these rules, one
can see that a simple microscope can be formed using two
lenses with different focal lengths (Fig. 3). The lens, L1,
with focal length, f1, images the sample plane and is called
the objective. The lens, L2, with focal length, f2, projects
the image onto the detector plane and is called the tube
lens. From simple geometry, two points P1 and P2
separated by x in the sample plane will be separated by
x(f2/f1) at the detector plane, where the ratio M f2/f1 is
called the magnification. One can see that the image in the
sample plane is enlarged by the magnification factor at the
detector.
By using the common wide-field fluorescence microscope
as an example, we can further examine the components of a
D
P2
P1
P1
f1
f1
f2
f2
P2
Microscope Optical Components

The optical principle underlying fluorescence microscopes
can be understood using basic ray tracing (7,8). The ray
tracing of light through an ideal lens can be formulated into
four rules: (1) A light ray originated from the focal point of a
lens will emerge parallel to the optical axis after the lens.
(2) A light ray propagating parallel to the optical axis will
pass through the focal point after the lens. (3) Light rays
L1
L2
Figure 3. The detection path of a microscope. Lenses L1 and L2

are the objective and the tube lens, respectively. L1 has focal
length f1 and L2 has focal length f2. For two points, P1 and P2
with separation, x, on the sample plane (S), these points are
projected to points P1 and P2 on the detector plane (D).
492
For an infinitely small emitter at the sample plane, the

image at the detector, the point spread function (PSF), is
not a single point. Instead, the intensity is distributed
according to an Airy function with a diameter, d:
L2
L2
BR
BR
L1
S
L4
L4
LS
1:22l
NA
(15)
DC
where M is the magnification of the system, l is the emission wavelength, and NA is the numerical aperture of the
objective, which is defined as (Fig. 2):
L1
NA n sin a
L3
LS
(a)
dM
(b)
Figure 4. Two configurations of fluorescence microscopy (a)

trans-illumination and (b) epi-illumination. The objective and
detection tube lenses are L1 and L2. The condenser is L3. The
excitation relay lenses are L4 and L40 . The sample and detector
planes are S and D respectively. The light source is LS. The
dichroic filter and the barrier filter are DC and BR.
complete fluorescence microscope system (Fig. 4a) (9). In

addition to the detection optical path, fluorescence microscope requires an excitation light source. The excitation
light source is typically placed in the focal point of a third
lens, L3. The lens collimates the excitation light and projects it uniformly on the specimen (Koehler illumination).
The lens, L3, is called the condenser. Since the excitation
light is typically much stronger than the fluorescence
emission, a bandpass filter is needed to block the excitation
light. In this trans-illumination configuration, it is often
difficult to select a bandpass filter with sufficient blocking
power without also losing a significant portion of the
fluorescence signal. To overcome this problem, an alternative geometry, epi-illumination, is commonly used
(Fig. 4b). In this geometry, lens L1 functions both as the
imaging objective and the condenser for the excitation
light. A couple of relay lenses (L4, L4) are used to focus
the excitation light at the back aperture plane of the
objective via a dichroic filter that reflects the excitation
light but transmits the fluorescence signal. The excitation
light is collimated by L1 and uniformly illuminates the
sample plane. The fluorescence signal from the sample is
collected by the objective and projected onto the detector
via the tube lens L2. Since the excitation light is not
directed at the detector, the task of rejecting excess excitation radiation at the detector is significantly easier. A
barrier filter is still needed to eliminate stray excitation
radiation from the optical surfaces.
From Fig. 2, one may assume that arbitrarily small
objects can be imaged by increasing the magnification
ratio. However, this is erroneous as the interference of
light imposes a resolution limit on an optical system (10).
The smallest scale features that can be resolved using
fluorescence microscopy are prescribed by the Abbe limit.
(16)
where a is the half-convergence angle of the light and n is

the index of refraction of the material between the lens and
the sample. Therefore, the images of two objects on the
sample plane will overlap if their separation is <1.22l/NA.
Since NA is always on the order of 1, an optical system can
only resolve two separate objects if their separation is on
the order of the wavelength of light.
Fluorescence Detectors and Signal Processing
Since the fluorescence signal is relatively weak, sensitive
detectors are crucial in the design of a high performance
fluorescence microscope. For a wide field microscope, the
most commonly used detectors are charged couple device
(CCD) cameras, which are area detectors that contain a
rectilinear array of pixels. Each pixel is a silicon semiconductor photosensor called a photodiode. When light is
incident upon an individual photodiode, electrons are generated in the semiconductor matrix. Electrodes are organized in the CCD camera such that the charges generated
by optical photons can be stored capacitatively during the
data acquisition. After data acquisition, manipulating
the voltages of the electrodes on the CCD chip allows
the charges stored in each pixel to be extracted from the
detector sequentially and read out by the signal conditioning circuit. These cameras are very efficient devices with a
quantum efficiency up to 80% (i.e., they can detect up to 8
out of 10 incident photons). Furthermore, CCD cameras
can be made very low noise such that even four to five
photons stored in a given pixel can be detected above
the inherent electronic noise background of the readout
electronics.
While CCDs are the detector of choice for wide-field
microscopy imaging, there are other microscope configurations (discussed below) where an array detector is not
necessary and significantly lower cost single element detectors can be used. Two commonly used single element
detectors are avalanche photodiodes (APDs) and photomultiplier tubes (PMTs).
Avalanche photodiodes and photomultiplier tubes have
been used in confocal and multiphoton microscopes. Avalanche photodiodes are similar to the photodiode element
in a CCD chip. By placing a high voltage across the device,
the photoelectron generated by the photon is accelerated
across the active area of the semiconductor and collide with
other electrons. Some of these electrons gain sufficient
mobility from the collision and are accelerated toward
the anode of the device themselves. This results in an
avalanche effect with a total electron gain on the order
of hundreds to thousands. A sizable photocurrent is generated for each input photon. A normal photodiode or a
CCD camera does not have single photon sensitivity
because the readout electronic noise is higher than the
single electron level. The gain in the avalanche photodiode
allows single photon detection. Photomultiplier tubes operate on a similar concept. A photomultiplier is not a solidstate device, but a vacuum tube where the photons impact
the cathode and generates a photoelectron using the photoelectric effect. The electron generated is accelerated by a
high voltage toward a second electrode, called a dynode.
The impact of the first electron results in the generation of
a cascade of new electrons that are then accelerated toward
the next dynode. A photomultiplier typically has 510
dynode stages. The electron current generated is collected
by the last electrode, the anode, and is extracted. The
electron gain of a photomultiplier is typically >110
million. While APDs and PMTs are similar devices, they
do have some fundamental differences. The APD are silicon
devices and have a very high quantum efficiency ( 80%)
from the visible to the near-IR spectral range. The PMT
photocathode material has a typical efficient of 20%, but
can reach 40% in the bluegreen spectral range. However, PMTs are not sensitive in the redIR range with
quantum efficiency dropping to a few percent. On the other
hand, PMT have significantly higher gain and better temporal resolution.
Advanced Fluorescence Microscopy Configurations
In addition to wide-field imaging, fluorescence microscopy
can be implemented in other more advanced configurations
to enable novel imaging modes. We will cover four other
particularly important configurations: wide-field deconvolution microscopy, confocal microscopy, two-photon microscopy, and total internal reflection microscopy.
Wide-Field Deconvolution Microscopy. Wide-field
microscopy is a versatile, low cost, and widely used technique. However, cells and tissues are inherently three
dimensional (3D). In a thick sample, the signals from
multiple sample planes are integrated to form the final
image. Since there is little correlation between the structures at different depths, the final image becomes fuzzy.
The need for 3D resolved imaging has long been recognized. The iterative deconvolution approach has worked
well for relatively thin specimen, such as in the imaging of
organelle structures in cultured cells (11) (Fig. 5). In terms
of instrument modifications, the main difference between
deconvolution microscopy and wide-field microscope is the
incorporation of an automated axial scanning stage allowing a 3D image stack to be acquired from the specimen. An
initial estimate of the 3D distribution of fluorophores is
convoluted with the known PSF of the optical system. The
resultant image is then compared with the measured 3D
experimental data. The differences allow a better guess of
the actual fluorophore distribution. This modified fluorophore distribution is then convoluted with the system PSF
again and allows another comparison with experimental
data. This process repeats until an acceptable difference
between the convoluted image and the experimental data
493
(a)
(b)
Z = 0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
Figure 5. A comparison between (a) normal wide-field images

and (b) deconvoluted images (11). Green fluorescent protein
labeled mitochondria of a cultured cell was imaged by a widefield fluorescence microscope as a 3D image stack. The image stack
is deconvoluted and the significantly improved result is shown.
The axial position of the image stack is shown below in units of
micrometers.
is achieved. The deconvolution process in a wide-field

fluorescence microscope belongs to the class of mathematical problems called ill-posed problems (1214). An illposed problem does not have a unique solution, but
depends on the selection of approach constraints to reach
a final solution. One should consider the deconvoluted
images only as the best estimate of the real physical
structure given the available data. Furthermore, deconvolution algorithm is computationally intensive and often
fails in thick specimens.
Confocal Fluorescence Microscopy. Confocal fluorescence microscopy is a powerful method that can obtain
3D resolved sections in thick specimens by completely
optical means (1518). The operation principle of confocal
microscopy is relatively straightforward. Consider the following confocal optical system in the transillumination
geometry (Fig. 6). Excitation light is first focused at an
excitation pinhole aperture. An excitation tube lens collimates the rays and projects them toward the condenser.
The excitation light is focused at the specimen. The emitted
light from the focal point is collected by the objective and
collimated by the emission tube lens. The collimated light
is subsequently refocused at the emission pinhole aperture.
The detector is placed behind the aperture. As it is clear in
the ray tracing illustration, the fluorescence signal produced at the specimen position defined by the excitation
LS
L3
EXA
L3 S L1
L2
EMA
Figure 6. The configuration of a simple confocal microscope. The

objective and the detection tube lenses are L1, L2. The light source
is LS. The excitation aperture placed in front of the light source is
EXA. The relay lens that images the excitation aperture and
projects the image of the pinhole onto the specimen (S) are L3
and L30 . The fluorescence emission from the focal point (red rays)
are projected onto the emission aperture (EMA) by L1 and L2. The
signal is transmitted through EMA and is detected by the detector
(D). Fluorescence generated outside the focal plane in the
specimen (blue rays) are defocused at EMA and are mostly blocked.
494
Figure 7. A comparison between confocal (a) and wide field (b)

imaging of a plasmacytoma cell labeled with fluorescent
antiendoplasmin that binds mainly to the endoplasmic reticulum.
In the wide-field image, it is not possible to determine whether the
central nucleic region contains endoplasmin end the structure of the
cisternae are unclear (19).
pinhole aperture is exactly transmitted through the conjugate pinhole in the emission light path. However, for a
fluorescence signal generated above or below the focal
plane, the light is defocused at the emission pinhole aperture and is largely rejected. Hence, a pair of conjugate
pinholes allows the selection of a 3D defined volume. One
can show that a confocal microscope can image structures
in 3D with a volume resolution of 0.1 fl. This system
achieves 3D resolution, at the cost of obtaining fluorescence
signal from only a single point in the specimen. It is
necessary to raster scan the excitation focus to cover a
3D volume. Confocal microscopy has been used extensively
to investigate microstructures in cells and in the imaging of
tissues (19) (Fig. 7).
Two-Photon Fluorescence Microscopy. A two-photon
microscope is an alternative to confocal microscopy for
the 3D imaging of thick specimens. Denk, Webb, and coworkers in 1990 introduced two-photon excitation microscopy (18,20). Fluorophores can be excited by the simultaneous absorption of two photons each having one-half of the
energy needed for the excitation transition. Since the twophoton excitation probability is significantly less than the
one-photon probability, two-photon excitation occurs only
at appreciable rates in regions of high temporal and spatial
photon concentration. The high spatial concentration of
photons can be achieved by focusing the laser beam with a
high numerical aperture objective to a diffraction-limited
spot. The high temporal concentration of photons is made
possible by the availability of high peak power pulsed
lasers (Fig. 8). Depth discrimination is the most important
feature of multiphoton microscopy. In the two-photon case,
>80% of the total fluorescence intensity comes from a 1 mm
thick region about the focal point for objectives with
numerical aperture of 1.25. For a 1.25 NA objective using
excitation wavelength of 960 nm, the typical point spread
function has a fwhm of 0.3 mm in the radial direction and
0.9 mm in the axial direction (Fig. 8). Two-photon microscopy has a number of advantages compared with confocal
imaging: (1) Since a two-photon microscope obtains 3D
resolution by limitation the region of excitation instead
of the region of detection as in a confocal system, photodamage of biological specimens is restricted to the focal
point. Since out-of-plane chromophores are not excited,
they are not subject to photobleaching. (2) Two-photon

excitation wavelengths are typically redshifted to about
twice the one-photon excitation wavelengths in the IR
spectral range. The absorption and scattering of the excitation light in thick biological specimens are reduced.
(3) The wide separation between the excitation and emission spectra ensures that the excitation light and Raman
scattering can be rejected without filtering out any of the
fluorescence photons. An excellent demonstration of the
ability of two-photon imaging for deep tissue imaging is in
the neurobiology area (21) (Fig. 9).
Total internal reflection microscopy. Confocal and twophoton microscopy can obtain 3D resolved images from
specimens up to a few hundred micrometers in thickness.
However, both types of microscopy are technically challenging, require expensive instrumentation, and only can
acquire data sequentially from single points. Total internal
reflection microscopy (TIRM) is an interesting alternative
if 3D-resolved information is only required at the bottom
surface of the specimen, such as the basal membrane of a
cell (2224). Total internal reflection occurs at an interface
between materials with distinct indices of refraction
(Fig. 10). If light ray is incident from a high index prism,
n2, toward the lower index region, n1, at an angle u, the
light will be completely reflected at the interface if u is >uc,
the critical angle.
sin uc
n1
n2
(17)
While the light is completely reflected at the interface, the

electric field intensity right above the interface is nonzero,
but decays exponentially into the low index medium. The
decay length of the electric field is on the order of tens to
hundreds of nanometers. Compared with other forms of 3D
resolved microscopy, TIRM allows the selection of the
thinnest optical section, but only at the lower surface of
the sample. While prism launch TIRM as described is
simpler to construct, the bulky prism complicates the
routine use of TIRM for cell biology studies. Instead,
ultrahigh numerical aperture objectives have been produced (1.451.6 N). Light rays focus at the back aperture
plane of the objective that are sufficiently off axis will
emerge collimated, but at an oblique angle. If a specimen
grown on a high index coverglass is placed upon the
objective, total internal reflection can occur at the specimen-coverglass interface if the oblique angle is sufficiently
large. This approach has been described as the objective
launch TIRM and has been very successful in the study of
exocytosis processes (23) (Fig. 11).
FLUORESCENT PROBES
Fluorescence microscopy has found many applications in
biomedicine. This wide acceptance is a direct result of the
availability of an ever growing set of fluorescence probes
designed to measure cell and tissue structure, metabolism,
signaling processes, gene expression, and protein distribution (25,26). The synthesis of fluorescent probes dates back
to 1856, when William Perkin made the first synthetic
495
Figure 8. Two-photon microscopy optically sections and produces a fluorescent signal originating
only from the focal point (a) the geometry of two-photon fluorescence. In traditional one-photon
excitation, fluorescence is generated throughout the double inverted cones (blue arrow). Two-photon
excitation generates fluorescence only at the focal point (red arrow). (b) The submicron PSF of twophoton excitation at 960 nm: The full-widths at half maximum (fwhm) are 0.3 mm radially and 0.9
mm axially. (c) An experimental visualization of the small excitation volume of two-photon
fluorescence. One- and two-photon excitation beams are focused by two objectives (equal
numerical aperture) onto a fluorescein solution. Fluorescence is generated all along the path in
the one-photon excitation case (blue arrow), whereas fluorescence is generated only in a 3D confined
focal spot for two-photon excitation (red arrow) The reduced excitation volume is thought to lead to
less photodamage. (Please see online version for color figure)
probe from coal tar dye. Thereafter, many more synthetic

dyes became available: pararosaniline, methyl violet,
malachite green, safranin O, methylene blue, and numerous azo dyes. While most of these early dyes are weakly fluorescent, more fluorescent ones based on the xanthene and
acridine heterocyclic ring systems soon became available.
Optical Factors in the Selection of Fluorescent Probes
Before providing a survey of the wide variety of fluorescent
probes, it is important to first discuss the optical properties
of fluorescent probes that are important for microscopic
imaging: extinction coefficient, quantum yield, fluorescent
lifetime, photobleaching rate, and spectral characteristics.
One of the most important characteristic of a fluorescent
probe is its extinction coefficient. Extinction coefficient, e,
measures the absorption probability of the excitation light
by the fluorophore. Consider excitation light is transmitted
through a solution containing fluorophore at concentration
c with a path length l. The light intensities before and after
the solution are I0 and I. The extinction coefficient can then
be defined by Beers law:

log10
I0
ecl
I
(18)
Fluorescent probes with high extinction coefficients can

be excited by lower incident light intensity allowing the use
of lowest cost light sources and reducing the background
noise of the images originated from scattered excitation
light.
Quantum yield, Q, measures the relative contributions
of the radiative versus nonradiative decay pathways. High
quantum efficiency maximizes the fluorescent signal for
each photon absorbed. The combination of probe extinction
coefficient and quantum efficiency quantifies the total
conversion efficiency of excitation light into fluorescent
signal.
While e and Q determines excitation light conversion
efficiency, the maximum rate of fluorescent photon generation also depends on the lifetime, t, of the probe. Since a
molecule that has been excited cannot be reexcited until it
returns to the ground state, fluorescent lifetime
496
Figure 9. Fluorescein dextran labeled blood vessels in the

primary vibrissa cortex of a living rat brain imaged using twophoton microscope down to a depth of >500 mm, which
demonstrates the ability of this technique to image deep into
tissue (21).
L2
BR
L1
P
Figure 10. The configuration of a total internal reflection

fluorescence microscope. L1 and L2 are objective and tube lens,
respectively. The barrier filter is BR and the detector is D. The
prism is P. The excitation light (green) is incident up the prism at
angle, u, greater than the critical angle. The excitation light is
totally internally reflected from the surface. A magnified view is
shown on the left. The evanescence electric field induced by the
excitation light above the prism surface decays exponentially and
only induces strong fluorescence signal for probes close to the
surface of the prism. Please see online version for color figure.
Figure 11. Epi-illuminated wide field (EPI) and total internal

reflection (TIR) microscopy of bovine chromaffin cells containing
secretory granules marked with GFP atrial naturetic protein (23).
Only the lower plane of the cells contributes to the fluorescence
signal in TIR set-up.
determines the rate at which a single probe molecule can be

recycled. In general, for fluorescent probes with equal e and
Q, fluorescent photon production rate is an inverse function of probe lifetime. Further intersystem cross-rate also
plays a role in determining photon generation rate. Since
the triplet state has a very long lifetime, probes with high
intersystem cross-rates are trapped in the triplet state with
a relatively lower photon generation rate.
Photobleaching rate measures the probability that a
probe will undergo an excited-state chemical reaction
and become nonfluorescent irreversibly. Therefore, the
photobleaching rate of a probe limits the maximum number of fluorescence photons that can be produced by a single
fluorophore. Photobleaching rates of fluorophores vary
greatly. For example, rhodamine can survive up to
100,000 excitation, fluorescein a few thousand, and tryptophan can only sustain a few excitation events. Photobleaching can be caused by a variety of processes.
Generally, it is the result of a photochemical reaction in
the excited state of the probe. For example, a common
bleaching pathway is the generation of a triplet state that
reacts with oxygen dissolved in solution to generate singlet
oxygen and an oxidized molecule incapable of undergoing
the same electronic transition as before.
Spectral properties are also important in probe selection
for a number of reasons. First, selecting fluorescent probes
with well-separated excitation and emission spectra allow
more efficient separation of the fluorescence signal from
the excitation light background. Second, fluorescent probes
should be selected to match the detector used in the microscope that may have very different sensitivity across the
spectral range. For example, most PMTs have maximum

efficiency in the green spectral range, but very low efficiency in the red. Therefore, green emitting probes are
often better matches for microscopes using PMTs as detectors. Third, probes with narrow emission spectra allow a
specimen to be simultaneously labeled with different colors
providing a method to analyze multiple biochemical components simultaneously in the specimen.
Classification of Fluorescent Probes
There is no completely concise and definitive way to classify
the great variety of fluorescent probes. A classification can
be made based on how the fluorophores are deployed in
biomedical microscopy: intrinsic probes, extrinsic probes,
and genetic expressible probes.
Intrinsic Probes. Intrinsic probes refer to the class of
endogenous fluorophores found in cells and tissues. Many
biological components, deoxyribonuclic acid such as (DNA),
proteins, and lipid membrane are weakly fluorescent. For
example, protein fluorescence is due to the presence of
amino acids: tryptophan, tyrosine, and phenylalanine.
Among them, tryptophan is the only member with marginal quantum yield for microscopic imaging. However,
fluorescent imaging based on tryptophan provides very
limited information due to the prevalence of this amino
acid in many proteins distributed throughout cellular systems and provides no specificity or contrast. The most
useful intrinsic probes for microscopy imaging are a number of enzymes and proteins, such as reduced pyridine
nucleotides [NAD(P)H], flavoproteins, and protoporphyrin
IX. Both NAD(P)H and favoproteins are present in the
cellular redox pathway. The NAD(P)H becomes highly
fluorescent when reduced, whereas flavoprotein becomes
fluorescent when oxidized, while their redox counterparts
are nonfluorescent. These enzymes thus provide a powerful
method to monitor cell and tissue metabolism. Protoporphyrin IX (PPIX) is a natural byproduct in the heme
production pathway that is highly fluorescent. Certain
cancer cells have been shown to have upregulate PPIX
production relative to normal tissue and may be useful in
the optical detection of cancer. Another class of important
intrinsic fluorophores includes elastin and collagen, which
resides in the extracellular matrix allowing structural
imaging of tissues. Finally, natural pigment molecules,
such as lipofuscin and melanin, are also fluorescent and
have been used in assaying aging in the ocular system and
malignancy in the dermal system respectively.
Extrinsic Probes. Many extrinsic fluorescent probes
have been created over the last century. A majority of
these extrinsic fluorophores are small aromatic organic
molecules (2528). Many probe families, such as
xanthenes, canines, Alexas, coumarines, and acrinides
have been created. These probes are designed to span
the emission spectrum from near UV to the near-IR range
with optimized optical properties. Since these molecules
have no intrinsic biological activity, they must be conjugated to biological molecules of interest, which may be
proteins or structure components, such as lipid molecules.
497
Most common linkages are through reactions to amine and

thiol residues. Reactions to amine are based on acylating
reactions to form carboxamides, sulfonamides, or thioureas. Targeting thiol residue in the cysteines of proteins can
be accomplished via iodoacetamides or maleimides. Other
approaches to conjugate fluorophores to biological components may be based on general purpose linker molecules,
such as biotin-avidin pairs or based on photoactivable
linkers. A particularly important class of fluorophores
conjugated proteins is fluorescent antibodies that allow
biologically specific labeling.
In addition to maximizing the fluorescent signal, the
greater challenge in the design of small molecular probes is
to provide environmental sensitivity. An important class of
environmentally sensitive probes distinguishes the hydrophilic versus hydrophobic environment and results in a
significant quantum yield change or spectral shift based on
solvent interaction. This class of probes includes DAPI,
laurdan, and ANS, which have been used to specifically
label DNA, measure membrane fluidity, and sense protein
folding states, respectively. Another important class of
environmentally sensitive probes senses intracellular ion
concentrations, such as pH, Ca2, Mg2, Zn2. The most
important members of this class of probes are calcium
concentration sensitive because of the importance of calcium as a secondary messenger. Changes in intracellular
calcium levels have been measured by using probes that
either show an intensity or a spectral response upon calcium binding. These probes are predominantly analogues
of calcium chelators. Members of the Fluo-3 series and
Rhod-2 series allow fast measurement of the calcium level
based upon intensity changes. More quantitative measurement can be based on the Fura-1 and Indo-1 series that are
ratiometric. These probes exhibit a shift in the excitation or
emission spectrum with the formation of isosbestic points
upon calcium binding. The intensity ratio between the
emission maxima and the isosbestic point allows a quantitative measurement of calcium concentration without
influence from the differential partitioning of the dyes
into cells.
Quantum dots belong to a new group of extrinsic probes
that are rapidly gaining acceptance for biomedical imaging
due to a number of their very unique characteristics (29
31). Quantum dots are semiconductor nanoparticles in the
size range of 26 nm. Photon absorption in the semiconductor results in the formation of an exciton (an electronhole pair). Semiconductor nanoparticles with diameters
below the Bohr radius exhibit strong quantum confinement
effect, which results in the quantization of their electronic
energy level. The quantization level is related to particle
size where smaller particles have a larger energy gap. The
radiative recombination of the exciton results in the emission of a fluorescence photon with energy corresponding to
the excitons quantized energy levels. The lifetime for the
recombination of the exciton is long, typically on the order
of a few tens of nanoseconds. Quantum dots have been
fabricated from IIVI (e.g., as CdSe, CdTe, CdS, and ZnSe)
and IIIV (e.g., InP and InAs) semiconductors. Due to the
compounds involved in the formation of these fluorescent
labels, toxicity studies have to be realized prior to any
experiments. Recent research works have been devoted
498
to the better manufacture of these semiconductor crystals

including methods to form a uniform crystalline core and to
produce a surface capping layer that enhances the biocompability of these compounds, prevents their aggregation, and can maximize their quantum efficiency.
Furthermore, coating the surface of quantum dots with
convenient functional groups, including common linkages,
such as silane or biotin, has been accomplished to facilitate
linkage to the biological molecules. Quantum dots are
unique in their broad absorption spectra, very narrow
(15 nm) emission spectrum, and extraordinary photostability. In fact, quantum dots have been shown to have
photobleaching rates orders of magnitude below that of
organic dyes. Quantum dots also have excellent extinction
coefficients and quantum yield. While there are significant
advantages in using quantum dots, they also have a number of limitations including their relative larger size compared with organic dyes and their lower fluorescence flux
due to their long lifetime. Quantum dots have been applied
for single receptor tracking on cell surface and for the
visualization of tissue structures, such as blood vessels.
Genetic Expressible Probes. The development of genetically expressible probes has been rapid over the last decade
(32). The most notable of these genetic probes is green
fluorescent protein, GFP (33). The GFP was isolated and
purified from the bioluminescent jellyfish Aequorea
Victoria. Fusion proteins can be created by inserting
GFP genes into an expression vector that carries a gene
coding for a protein of interest. This provides a completely
noninvasive procedure and perfectly molecular specific
approach to track the expression, distribution, and trafficking of specific proteins in cells and tissues. In order to
better understand protein signaling processes and protein
protein interactions, fluorescent proteins of different colors
have been created based on random mutation processes.
Today, fluorescent proteins with emission spanning the
spectral range from blue to red are readily available.
Expressible fluorescent proteins that are sensitive to cellular biochemical environment, such as pH and calcium,
have also been developed. Novel fluorescent proteins with
optically controllable fluorescent properties, such as photoactivatable fluorescent proteins, PA-GFP, photoswitchable
CFP, and pKindling red have been created and may be used
in tracing cell movement or protein transport. Finally,
proteinprotein interactions have been detected based on
a novel fluorescent protein approach in which each of the
interacting protein pairs carries one-half of a fluorescent
protein structure that is not fluorescent. Upon binding of
the protein pairs, the two halves of the fluorescent protein
also recombine, which results in a fluorescent signal.
ADVANCED FUNCTIONAL IMAGING MODALITIES AND

THEIR APPLICATIONS
A number of functional imaging modalities based on fluorescent microscopy have been developed. These techniques
are extremely versatile and have found applications ranging from single molecular studies to tissue level experiments. The implementation of the most common imaging
modalities will be discussed with representative examples

from the literature.
Intensity Measurements
The most basic application of fluorescence microscopy consists in mapping fluorophore distribution based on their
emission intensity as a function of position. However, this
map is not static. Measuring intensity distribution as a
function of time allows one to follow the evolution of
biological processes. The fastest wide-field detectors can
have a frame rate in the tens of kilohertz range, unfortunately at the expense of sensitivity and spatial resolution.
They are used to study extremely fast dynamics, such as
membrane potential imaging in neurons. For 3D imaging,
point scanning techniques are typically slower than widefield imaging, but can reach video rate speed using multifoci illumination.
Dynamic intensity imaging has been used at the tissue
level to follow cancer cells as they flow through blood
vessels and extravasate to form metastases, or in embryos
to track the expression of a regulatory protein at different
developmental stages. One commonly used technique to
follow the movements of protein in cellular systems is
fluorescent recovery after photobleaching (FRAP). In
FRAP studies, a small area of a cell expressing a fluorescently labeled protein is subjected to an intense illumination that photobleaches the dye and leads to a drastic drop
in fluorescence intensity. The rate at which the intensity
recovers provides a measure of the mobility of the protein of
interest.
An important aspect of the fluorescent microscopy technique lies also in the image analysis. Particle tracking
experiments are an excellent example. Zuhang and coworkers (34) studied the infection pathway of the influenza
virus labeled with a lipophilic dye in CHO cells. Each frame
of the movie recorded was analyzed to extract the position
of the virus particles with 40 nm accuracy. Three different
stages of transport after endocytosis of the virus particle
were separated, each involving different transport
mechanisms transduced by a different protein as shown
on Fig. 12. The first stage is dependant on actin and results
in an average transport distance of 2 mm from the initial
site of binding at the cell periphery. The second stage is
characterized by a sudden directed displacement that
brings the virus close to the nucleus with a speed of 14
mm s1 that is consistent with the velocity of dynein motors
on microtublues. The last stage consists of back and forth
motion in the perinuclear region. This is followed by the
fusion of the endosome with the virus and the liberation of
the genetic material. This event is identified by a sudden
increase in the fluorescence intensity due to the dequenching of the fluorescent tags on the virus.
Spectral Measurements
An extremely important feature of fluorescent microscopy
is the ability to image many different fluorescent species
based on their distinct emission spectra. Dichroic bandpass
filters optimized for the dyes used in the experiment can
discriminate efficiently between up to four or five different
fluorophores.
499
Figure 12. Particle tracking of virus

infecting a cell. (a) Trajectory of the
virus. The color of the trajectory codes
time from 0 s (black) to 500 s (yellow).
The star indicates the fusion site of the
virus membrane with the vesicle. (b)
Time trajectories of the velocity
(black) and fluorescence (blue) of the
virus particle (34). Please see online
version for color figure.
In a study of connexin trafficking, Ellisman and coworkers (35) used a dual labeling scheme to highlight
the dynamics of these proteins. Using a recombinant protein fused to a tetracystein receptor domain, the connexin
was stably labeled with a biarsenical derivate of fluorescein
or resorufin (a red fluorophore). The cells expressing these
modified proteins were first stained with the green fluorophore and incubated 48 h. The proteins produced during
this incubation period are fluorescently tagged in a second
staining step with the red fluorophore. The two-color
images highlight the dynamics of the connexin refurbishing at the gap junction. As shown on Fig. 13, the older
proteins are found in the center and are surrounded by the
newer proteins.
For wide-field fluorescence imaging using a CCD camera, spectral information is collected sequentially while
position information is collected at once. Bandpass filters
can be inserted to select the emission wavelength in
between image frames. This procedure is relatively slow
and can result in image misregistration due to the slight
misalignment of the filters. This problem can be overcome
by the use of electronically tunable filter. Two types of
electronically tunable filters are available based either on
liquid-crystal technology or on electrooptical crystals.
Liquid-crystal tunable filters are made of stacks of birefringent liquid-crystal layers sandwiched between polarizing filters. Polarized light is incident upon the device. The
application of a voltage on the liquid-crystal layer produces
a wavelength dependent rotation of the polarization of light
as the light is transmitted through the liquid-crystal
layers. After cumulative rotations through the multiple
layers, only the light at a specific spectral range is at the
correct polarization to pass through the final polarizer
without attenuation. The second type is acoustooptics tunable filters (AOTFs). An AOTF works by setting acoustic
vibration at radio frequency (rf) through an electrooptical
crystal to create a diffraction grating that singles out the
appropriate wavelength with a few nanometer bandwidth.
The main advantage of AOTF is that the wavelength
selection is realized by tuning the acoustic wave frequency,
which can be done in a fraction of a millisecond while the
liquid-crystal tunable filters operate with a time constant
of hundreds of milliseconds. The latter, however, have a
larger clear aperture and selectable bandwidth ranging
from a fraction of a nanometer up to tens of a nanometer.

Liquid-crystal filters are more often used for emission
filtering while the acoustooptic filters are more commonly
used for excitation wavelength selection.
Typical emission spectra from molecular fluorophores
have a sharp edge at the blue end of the spectrum, but have
a long tail extending far into the red due to electronic
relaxation from the excited state into a vibrationally
excited ground state. When imaging with a few color
channels, where each channel represents a single chromophore, one has to be careful to take into account the
spectral bleedthrough of each dye into the neighboring
channels. Collecting signal in a larger number of channels
allows the use of a linear unmixing technique to account for
the real shape of the emission spectra of each dye and
accounts more precisely for their contributions in each
pixel of the image. This technique can be implemented
using tunable filters with a narrow bandwidth and CCD
camera detectors. It has also been shown that an interferometer can be used to encode the spectral information in
the image on the CCD camera. An image is then recorded
for each step of the interferometer and a Fourier transform
analysis allows the recovery of the spectral information.
Although it requires more advanced postprocessing of the
image data, this approach offers a large spectral range and
a variable spectral resolution unmatched by the tunable
filters.
Figure 13. Connexin trafficking at gap junction. The newly

produced proteins are labeled in red after 4 h (a and b) or 8 h (c
and d) hours after the first staining step with green. The older
proteins occupy the periphery of the gap junction, while the new
ones are localized in its center (35). Please see online version for
color figure.
500
In scanning systems, such as confocal microscopes, the

use of dichroic beamsplitters can be readily constructed to
simultaneously resolve two or three spectral channels in
parallel at each scan position. If more spectral channels are
desired for spectral decomposition measurement, the emission can be resolved in a multichannel detector using a
grating or a prism to separate the different wavelength
components. This has been used to separate the contribution of dyes with very similar emission spectra like GFP
and fluorescein, or resolve the different intrinsic fluorophores contained in the skin where many fluorophores with
overlapping spectra are present.
A particularly promising class of probes for spectral
imaging are quantum dots. As discussed previously, the
emission spectra of quantum dots are very narrow and can
be tuned by changing their size. Further, all quantum dots
have a very broad excitation spectrum and a single excitation wavelength larger than the band gap energy can lead
to the efficient excitation of many different colored quantum dots simultaneously. In their report, Simon and coworkers (36) used these particles to track metastatic cells
injected in the tail of a mouse as they extravasate into lung
tissue. Using spectrally resolved measurements, they
demonstrate their ability to recognize at least five different
cell populations each labeled with different quantum dots.
Figure 14 shows an image of cells labeled with different
quantum dots and the emission spectra from each of these
particles. The difference in emission spectra allows an easy
identification of each cell population.
Lifetime Resolved Microscopy
Measurement of the fluorescence lifetime in a microscope
provides another type of contrast mechanism and can be
used to discriminate dyes emitting in the same wavelength
range. It is also commonly used to monitor changes in the
local environment of a probe measuring the pH or the
concentration of cations In situ. The fluorescence lifetime
can be shortened by interaction of the probe with a
quencher, such as oxygen. Another type of quenching is
induced by the presence of the transition dipole of other
dyes, which are in close vicinity and lifetime measurements can be used to quantify energy-transfer processes
(discussed further in a later section).
There are two methods to measure the fluorescence

lifetime in a microscope. One is in the time domain and
the other is in the frequency domain. In the time domain, a
light pulse of short duration excites the sample and the
decay of the emission is timed. The resulting intensity
distribution is a convolution between the instrument
response and the exponential decay of the fluorophore.

Z t
T
It I0
Gt T exp
dT
(19)
t
0
In the frequency domain, the excitation light is modulated at
frequency v. The intrinsic response time of the fluorescence
acts as a low pass filter and the emitted signal is phase
shifted and demodulated. Both the demodulation and the
phase shift can be linked to the fluorescence lifetime.
Df a tanvt
(20)
1
M p
1 v2 t 2
(21)
In order to obtain a measurable phase shift and modulation, the frequency has to be on the same order of
magnitude as the lifetime (i.e., 108 Hz). However, it is
difficult to measure these two parameters at such high
frequencies. Therefore, one typically uses a heterodyne
detection to lower the frequency to the kilohertz range by
modulating the detector at a frequency close to the
excitation frequency.
For wide-field microscopy, an image intensifier is placed
in front of the CCD camera to modulate the gain of detection.
In the time domain, a short time gate is generated to collect
the emission at various times after the excitation. In the
frequency domain, the image intensifier is modulated at
high frequencies and a series of images at different phases
are acquired. In laser scanning confocal and multiphoton
microscopes, time correlated single-photon counting is the
method of choice for lifetime measurements in the time
domain because it offers an excellent signal/noise ratio at
low light levels. Every time a photon is detected, the time
elapsed since the excitation of the sample is measured. A
histogram of all the times of arrival yields a decay curve of
the fluorescence in each pixel of the image. For brighter
Figure 14. Spectral imaging of cells labeled by quantum

dots. Cells were labeled with five different quantum dots
and imaged in a multiphoton microscope. Each symbol
represents a different quantum dot. The symbols on the
image match the emission spectra seen on the graph. The
spectral imaging set-up allows to discriminate between
the different cell populations (36).
501
Figure 15. Quantification of the pH

of the skin by lifetime imaging. (a)
Intensity, (b) modulation, (c) lifetime,
and (d) pH maps of a mouse skin at
different
depth.
The
lifetime
measurements allow a determination
of the pH independently of the
intensity changes recorded between
different imaging depth (37).
samples, a frequency domain approach using modulated

detectors can also be used to measure the lifetime.
To measure the pH in the skin of a mouse, Clegg and coworkers (37) used a modified fluorescein probe whose lifetime varies from 2.75 ns at pH 4.5 to 3.9 ns at pH 8.5. As
they image deeper in the skin, they observe that the
average pH increases from 6.4 at the surface up to >7 at
15 mm depth. The extracellular space is mostly acidic (pH
6), while the intracellular space is at neutral pH. Typically,
pH is measured in solution by correlating fluorescence
intensities with specific pH levels. This approach is not
suitable for tissues, as in the skin, since the dye is unevenly
distributed throughout the tissue (Fig. 15) due to differential partitioning. A measurement of pH based on fluorescence lifetime is not dependent on probe concentration
and thus the pH can be measured in the intra and extracellular space at various depths in the skin.
Polarization Microscopy
Polarization microscopy is a technique that provides information about the orientation or the rotation of fluorophores. Linearly polarized excitation light results in
preferential excitation of molecules with their transition
dipole aligned along the polarization. If the molecule is in a
rigid environment, the emitted fluorescence will mostly
retain a polarization parallel to the excitation light. However, if the molecule has time to rotate before it emits a
photon, this will randomize the emission polarization. The
anisotropy r is a ratio calculated from the intensity parallel
Ik and perpendicular I? to the incident polarization and is a
measure of the ability of the molecule to rotate.
r
Ik I ?
Ik 2I ?
(22)
This ratio is mostly governed by two factors, which are

the fluorescence lifetime t and the rotational correlation
time u.
r
r0
1 t=u
(23)
where r0 is the fundamental anisotropy. Molecules with a

short fluorescence lifetime and a long rotational correlation time (t < u) will have a high anisotropy. In the
opposite case, where molecules can freely rotate during
the time they reside in the excited state, the anisotropy
will be low. An approximate measurement of the mobility
of a molecule can be obtained by exciting the sample at
different polarization angles. A proper measurement of
the anisotropy requires both a linearly polarized excitation light source and the detection of the parallel and
perpendicular component of the fluorescence light using a
polarizer. This technique has been used to measure
viscosity and membrane fluidity In vivo. It has been
applied to quantify enzyme kinetics, relying on the fact
that the cleavage of a fluorescently labeled substrate leads
to a faster tumbling and thus a decrease in anisotropy.
Goldstein and co-workers (38) used polarization microscopy at the single-molecule level to study the orientation of
the kinesin motor on a microtubule. A thiol reactive rhodamine dye was attached to cysteines on the motor protein.
Microtubules decorated with the modified kinesin were
imaged under a different polarization angle. In the presence
of adenosine monophosphate (AMP)(PNP) [a nonhydrolyable analogue of adenonine triphosphats (ATP)], the fluorescence intensity depends strongly on the angle of
polarization of the excitation light (Fig. 16) proving that
the kinesin maintains a fixed orientation. In the presence of
adenonine5diphosphate (ADP), however, the anisotropy is
lower (no dependence on excitation polarization angle),
502
Figure 16. Mobility of single kinesin motors on microtubules

probed by polarization microscopy. (a) Image of microtubules
sparsely decorated with kinesin motors in presence of AMP
PNP and ADP. (b) Time course of the fluorescent intensity
recorded from single molecule excited with linearly polarized
light at four different angles. The large fluctuations of the
fluorescence intensity as function of the excitation polarization
in the AMPPNP case demonstrate the rigidity of the kinesin
motor on the microtubule (38).
leading to the conclusion that the kinesin is very mobile,

while still attached to the microtubule.
Fluorescence Resonance Energy Transfer
Fo rster resonance energy transfer (FRET) is a technique
used to monitor interaction between two fluorophores on
the nanometer scale. When a dye is promoted to its excited
state, it can transfer this electronic excitation by dipole
dipole interaction to a nearby molecule. Due to the nature
of this interaction, Fo rster predicted a dependence of the
FRET efficiency on the sixth power of distance that was
demonstrated experimentally by Stryer with a linear polypeptide of varying length (39). The efficiency E of the
process varies as function of the distance, R, between
the two molecules as
E
R60
R60
R6
Most FRET experiments are based on the measurement

of the intensity of the donor and of the acceptor because the
presence of FRET in a system is characterized by a
decrease in the emission of the donor and an increase in
the acceptor signal. Thus, in principle, a two color channel
microscope is sufficient to follow these changes. However,
experimental artifacts, such as concentration fluctuation
and spectral bleed, complicate the analysis of these images
and many different correction algorithms have been
developed.
FRET measurements have been used in molecular studies to measure distances and observe dynamic conformational changes in proteins and ribonuclic acid (RNA). In
cellular studies, FRET is often used to map protein interactions. By labeling one protein with a donor dye and its
ligand with an acceptor dye, energy transfer will occur only
when the two proteins are bound such that the dyes come in
close proximity of each other.
The addition of fluorescence lifetime imaging provides
the additional capability of retrieving the proportion of
fluorophore undergoing FRET in each pixel of an image
independently of concentration variations. This is possible
because the fluorescence lifetime of a FRET construct is
shorter than the natural decay of the dye. Thus if one has a
mixture of interacting and free protein, fitting a double
exponential to the fluorescence decay allows us to retrieve
the proportion of interacting protein. This has been applied
by Bastiaens and co-workers (40) to the study the phosphorylation of the EGF receptor ErB1. The transmembrane receptor is fused with a GFP protein. The
phosphrylation of the protein is sensed by an antibody
labeled with a red dye (Cy3). When the Erb1 is phosphorylated, the antibody binds to the protein and FRET occurs
due to the short distance between the antibody and the
GFP. The ErB1 receptors can be stimulated by EGF coated
beads leading to phosphorylation and FRET. The time
course of the stimulation is followed and for each cell
and the fraction of phosphorylated receptors at various
time interval is shown in Fig. 17. After 30 s, the FRET
events are localized at discrete locations. But after 1 min,
the whole periphery of the cell displays high FRET, demonstrating the lateral signaling of the receptor after activation at discrete location by EGF coated beads.
(24)
Where R0 is called the Fo rster distance, which depends on

Avogadros number NA, the index of refraction of the
medium n, the quantum yield of the donor molecule QD,
the orientation factor k, and the overlap integral J.
R60
9000 ln10k2 QD
J
128p5 NA n4
(25)
k represents the relative orientation of the transition dipole

of the donor and acceptor molecules. In most cases, a
random interaction is presumed and this factor is set to
two-thirds. The overlap integral J represents the energy
overlap between the emission of the donor and the absorption of the acceptor. For well-matched fluorophore pairs, R0
is on the order of 47 nm.
Figure 17. Time course of the phosphorylation of EGF receptors

ERB1 after stimulation by EGF coated beads observed by FRET
between a GFP modified EGF receptor and a phosphorylation
antibody labeled with Cy3. While the GFP intensity is remains
relatively constant, the concentration of the Cy3 tagged antibody
clearly increases after the stimulation. This leads to an increase
FRET signal as function of time (40).
CONCLUSION
The utility of fluorescence microscopy lies in its ability to
study biological structure and function In vivo. The exquisite sensitivity and image contrast of fluorescence microscopy allow biological structures to be imaged on the
submicron length scale. The greatest power of fluorescence
microscopy lies in its ability to determine biochemical
functions using assays based on fluorescence spectroscopy.
With the availability of more versatile instruments, more
fluorophores unit greater molecular and environmental
specificity, the impact of fluorescence microscopy technology on biomedical science will only increase.
BIBLIOGRAPHY
1. Wang XF, Herman B. Fluorescence Imaging Spectroscopy
and Microscopy. New York: Wiley; 1996.
2. Inoue S, Spring KR. Video Microscopy: the Fundamentals.
3. Herman B. Fluorescence Microscopy. Oxford: Bios Scientific
Publishers / Springer in Association with the Royal Microscopy Society; 1998.
4. Cantor CR, Schimmel PR. Biophysical Chemistry. San Francisco: Freeman; 1980.
5. Valeur B. Molecular Fluorescence: Principles and Applications. Weinheim; New York: Wiley-VCH; 2002.
6. Lakowicz JR. NetLibrary Inc. Topics in Fluorescence Spectroscopy. Kluwer Academic; 2002.
7. Klein MV, Furtak TE. Optics. New York: Wiley; 1986.
8. Hecht E. Optics. Reading, (MA): Addison-Wesley; 2002.
9. Gu M. Advanced Optical Imaging Theory. Berlin: New York:
Springer; 2000.
10. Born M,Wolf E. Principles of Optics: Electromagnetic Theory
of Propagation, Interference and Diffraction of Light. Cambridge: New York: Cambridge University Press; 1999.
11. Rizzuto R, Carrington W, Tuft RA. Digital imaging microscopy of living cells. Trends Cell Biol 1998;8:288292.
12. Agard DA, Hiraoka Y, Shaw P, Sedat JW. Fluorescence microscopy in three dimensions. Methods Cell Biol 1989;30:353377.
13. Carrington WA, et al. Superresolution three-dimensional
images of fluorescence in cells with minimal light exposure.
Science 1995;268:14831487.
14. Krishnamurthi V, Liu YH, Bhattacharyya S, Turner JN,
Holmes TJ. Blind Deconvolution Of Fluorescence Micrographs By Maximum-Likelihood-Estimation. Appl Opt
1995;34:66336647.
15. Wilson T, Sheppard CJR. Theory and Practice of Scanning
Optical Microscopy. New York: Academic; 1984.
16. Pawley JB. Handbook of Confocal Microscopy. New York:
Plenum; 1995.
17. Gu M. Principles of Three-Dimensional Imaging in Confocal
Microscopy. Singapore: World Scientific; 1996.
18. MastersBR. Selected Papers on Multiphoton Excitation Microscopy. Bellingham: SPIE Optical Engineering Press; 2003.
19. Amos WB, White JG. How the confocal laser scanning microscope entered biological research. Biol Cell 2003;95:335342.
20. Denk W, Strickler JH, Webb WW. Two-photon laser scanning
fluorescence microscopy. Science 1990;248:7376.
21. Kleinfeld D, Mitra PP, Helmchen F, Denk W. Fluctuations
and stimulusinduced changes in blood flow observed in individual capillaries in layers 2 through 4 of rat neocortex. Proc
Natl Acad Sci U. S. A. 1998;95:1574115746.
22. Axelrod D. Total internal reflection fluorescence. Ann Rev
Biophys Bioeng 1984;13:247268.
503
23. Axelrod D. Total internal reflection fluorescence microscopy

in cell biology. Traffic 2001;2:764774.
24. Axelrod D. Selective imaging of surface fluorescence with
very high aperture microscope objectives. J Biomed Opt
2001;6:613.
25. Mason WT. Fluorescent and Luminescent Probes for Biological Activity: a Practical Guide to Technology for Quantitative Real-time Analysis. San Diego: Academic; 1999.
26. Slavic J. Fluorescence Microscopy and Fluorescent Probes.
27. Tsien RY. Fluorescent probes of cell signaling. Annu Rev
Neurosci 1989;12:227253.
28. Tsien RY. Fluorescent indicators of ion concentrations. Methods Cell Biol 1989;30:127156.
29. Bruchez Jr M, Moronne M, Gin P, Weiss S, Alivisatos AP.
Semiconductor nanocrystals as fluorescent biological labels.
Science 1998;281:20132016.
30. Chan WC, et al. Luminescent quantum dots for multiplexed
biological detection and imaging. Curr Opin Biotechnol
2002;13:4046.
31. Michalet X, et al. Quantum dots for live cells, in vivo imaging,
and diagnostics. Science 2005;307:538544.
32. Zhang J, Campbell RE, Ting AY, Tsien RY. Creating new
fluorescent probes for cell biology. Nat Rev Mol Cell Biol
2002;3:906918.
33. Chalfie M, et al. Green Fluorescent Protein as a Marker for
Gene Expression. Science 1994;263:802805.
34. Lakadamyali M, Rust MJ, Babcock HP, Zhuang X. Visualizing infection of individual influenza viruses. Proc Natl Acad
Sci U. S. A. 2003;100:92809285.
35. Gaietta G, et al. Multicolor and electron microscopic imaging
of connexin trafficking. Science 2002;296:503507.
36. Voura EB, Jaiswal JK, Mattoussi H, Simon SM. Tracking
metastatic tumor cell extravasation with quantum dot nanocrystals and fluorescence emissionscanning microscopy. Nat
Med 2004;10:993998.
37. Hanson KM, et al. Two-photon fluorescence lifetime imaging
of the skin stratum corneum pH gradient. Biophys J
2002;83:16821690.
38. Sosa H, Peterman EJ, Moerner WE, Goldstein LS. ADPinduced rocking of the kinesin motor domain revealed by
single-molecule fluorescence polarization microscopy. Nat
Struct Biol 2001;8:540544.
39. Stryer L, Haugland RP. Energy Transfer: a spectroscopic
ruler. Proc Natl Acad Sci U. S. A. 1967;58:712726.
40. Verveer PJ, Wouters FS, Reynolds AR, Bastiaens PIH. Quantitative imaging of lateral ErbB1 receptor signal propagation
in the plasma membrane. Science 2000;290:15671570.
See also FLUORESCENCE
MEASUREMENTS; MICROSCOPY, CONFOCAL

EWA P. WOJCIKIEWICZ
KWANJ JOO KWAK
VINCENT T. MOY
University of Miami
Miami, Florida
INTRODUCTION
Recent advances in technology have allowed us to study our
world at molecular, even subatomic, resolution. One of the
devices in the forefront of such studies is the atomic force
microscope (AFM), which is a relatively complex device
504
with two major applications. It can be used as an imaging

device, which allows for the acquisition of atomic-level
images of biological structures as well as to measure forces
of interactions between two opposing surfaces down to the
single-molecule level.
Imaging AFM
The AFM was originally designed as an imaging tool (1). It
was modified from the design of the scanning tunneling
microscope (STM). The AFM acquires topographic images
by methodically scanning a sample with a flexible probe,
called a cantilever, which bends according to the contours
of the samples surface. The bending of the cantilever is
translated into an image map, which reveals the height
differences in the surface being scanned. It is possible to
image biological samples under physiological conditions as
imaging can be done in both air and liquid. The resulting
resolution of such maps is at the atomic level (2,3).
The imaging AFM has been used to image many biological samples ranging from genetic material to cells to
bone. A few of these studies will be highlighted. One of
the earliest biological materials to be imaged was DNA,
which has been imaged in many forms to date, including
double- and single-stranded forms as well as more complex
structures. The AFM has also been used for many applications including DNA sizing, previously only achieved using
gel electrophoresis, DNA mapping, hybridization studies,
and examinations of protein-DNA interactions (4). AFM
studies of RNA were also conducted. Unlike DNA, which
mainly forms a double-helical structure, RNA has the
ability to form more advanced structures that do not rely
solely on WatsonCrick base pairing. One example are the
so-called kissing-loop structures imaged by Hansma et al.
(4) (Fig. 1). Not only was the AFM used in imaging of such
structures, many of them 3D, but also played an important
role in designing them (5). Unlike other imaging techniques, AFM sudies can be done under physiological conditions allowing for the imaging of biological processes.
Images of transcription complexes have been obtained,
for example, E.coli RNA polymerase in complex with
DNA. These studies are the only of their kind that can
answer certain specific questions as to how the RNA transcription process takes place. One is able to visualize how
the DNA does not get entangled in the nascent RNA
strands. Such studies detailing the structure-function relationship of the transcription process are key in furthering
our understanding of gene expression (6,7).
Also, imaging of cells was conducted to examine the
structure of the cellular cortex in detail. The cell cytoskeleton is known to be involved in affecting cell shape as well
as movement and other cellular responses to biochemical
and biophysical signals. At present, relatively little is
known about the mechanical organization of cells at a
subcellular level. Pesen et al. (8) studied the cell cortex
of bovine pulmonary artery endothelial cells (BPAECs)
using AFM and confocal fluorescence microscopy (CFM).
They were able to identify a coarse and fine mesh that
make up the cortical cytoskeleton. These two types of mesh
appear to be intertwined (Fig. 2) (8). Such details are not
distinguished in imaging studies using fixed cells.
Figure 1. Kissing-loop RNA structures. (a,c) Kissing-loop RNAs

at low concentrations. The three arms of the individual RNAs are
visible. (B) Kissing-loop RNAs at a concentration 10-fold higher
than in a and c. The individual structures are less well-defined.
Scale bars 100 nm for all images.
Other imaging studies have looked at tendon and bone,

both of which are composed of type I tropo-collagen, which
was done by acquisition of high resolution AFM images of
type I collagen in conjunction with force spectroscopy
studies, namely protein unfolding, which is described in
the following section. Figure 3 reveals these high resolution collagen type I images. They were acquired using two
different concentrations of collagen, which resulted in
Figure 2. AFM Images of the cortical mesh of bovine pulmonary

artery endothelial cells (BPAEC). (ad) High magnification
deflection AFM images of the cortical mesh of living BPAECs in
a physiological saline. The filamentous mesh appears to be
organized on two length scales, with coarse mesh (arrowheads)
and fine mesh filaments (arrows). The two meshes are likely to be
intertwined, although it is possible that the fine mesh is layered
over the coarse mesh. Lateral resolution in these images is
125 nm.
slightly different orientations of collagen: random at the

lower concentration (Fig. 3a) and oriented unidirectionally
in the higher concentration (Fig. 3b). In these studies, the
AFM was used to investigate the mechanical properties of
this collagenous tissue, which are altered in diseases such
as osteoporosis. Being familiar with such properties is
important for gaining further understanding as well as
preventing and curing bone diseases (9).
Force Spectroscopy
The AFM can also be operated in the force scan mode,
which allows for the measurement of adhesion forces
Figure 3. Topographical images (height; tapping mode in air) of

type I collagen monomers on a mica substrate. (a) low (1 mg/ml)
concentration of collagen, (b) high (10 mg/ml) concentration of
collagen.
505
between receptors and their corresponding binding partners, or ligands. In studies of ligand-receptor forces, the
receptor is immobilized on the surface of a flexible AFM
cantilever whereas the ligand is attached to a suitable
substrate. The deflection of the cantilever during the
approach and withdrawal of the cantilever from the substrate allow for the force of the interaction to be measured.
These type of experiments provide information simulating
the influence of internal and external forces that these
receptors would experience in the body, for example, the
shear stress experienced by a blood cell attached to the
endothelium while blood rushes past it. Such information
was previously unavailable when receptor-ligand interactions were examined using more traditional biochemical
techniques. AFM has made it possible to acquire measurements that reveal the mechanical properties of biomolecules under applied force. Measurements of the unbinding
force of a single ligand-receptor interaction can now be
acquired with the AFM (1012).
The AFM can also be used in adhesion studies involving
whole cells (13,14). In these studies, the interaction
between a cell expressing a particular receptor of interest
and its ligand protein or another cell expressing the ligand
is measured. The cell adhesion experiments allow for the
acquisition of both single-molecule measurements, like in
the above-mentioned studies, as well as multiple-bond
interactions. The advantage of using the AFM in cell
adhesion studies is the high specificity and wealth of
information that is obtained. The AFM force scans provide
information about the individual bond strengths as well as
the force and work that is required to separate the entire
complex. Combining single-molecule and multiple-bond
data allows us to describe the thermodynamic model of
the separation of a particular complex in addition to the
mechanism of its action on the cellular scale (15,16).
The AFM can also serve as a microindenter that probes
soft samples, including cells revealing information about
their mechanical properties. The mechanical properties of
cells play an important role in such essential physiological
processes such as cell migration and cell division. Understanding these properties can later help scientists to identify when certain processes may be taking place. The
mechanical properties of cells are chiefly determined by
their actin cytoskleleton, which is the cells backbone.
This type of information, which cannot be obtained using
standard cell biology methods, allows for the estimation of
the Youngs modulus of living cells (16). The Youngs
modulus is a calculated value, which provides information
regarding the compliance or elasticity of a cell. Such
experiments may be done with either the imaging AFM
or using force spectroscopy. Manfred Radmachers group
has conducted such measurements with an imaging AFM
in force mapping mode. The advantage of such measurements is the wealth of information that they provide. These
experiments reveal not only the elastic properties of the
cells being examined but also topographical information.
They can also be performed in conjunction with video
microscopy to further confirm what one is visualizing
and that cells are not undergoing damage (17,18). In force
spectroscopy experiments, the Youngs modulus is
obtained by poking the cell cantilever tip. This type of
506
information can be correlated with adhesion data to determine whether elasticity changes in response to drugs or
other stimulation have an effect on the strength of cell
adhesion (19).
Another application allows scientists to study protein
folding. Proteins are composed of amino acid chains, which
constitute the primary protein structure. These amino acid
chains have to undergo folding into tertiary and secondary,
3D structures, which is essential for the proper functioning
of these proteins. Recently, advances in AFM have made it
possible to study this fascinating biological process and
bring new insight into the energy landscapes of protein
folding. Proteins involved in mechanical functions are
composed of multiple domains, which fold individually.
One of these domains is fibronectin. The most common
form of fibronectin is fibronectin type III (FN-III), which is
found in an estimated 2% of all animal proteins and has
thus been studied extensively. FN-III domains are found in
the muscle protein titin, which is responsible for the passive elasticity of muscle. These domains have been found to
unravel during forced extension. Understanding the forces
required in unfolding events as well as the time it takes for
unfolding to happen can be critical for the physiological
functions of mechanical proteins such as titin (9,2022).
Now that a brief overview of the potential applications of
the AFM has been provided, it is important to understand
the principles behind its operation. The following section
focuses on the theory of data acquisition using AFM as well
as descriptions of the equipment itself. The last section of
this article provides a more in-depth evaluation of the
technique in addition to discussing the most recent
advances in the field.
THEORY AND EXPERIMENTAL APPROACH
This section focuses on the theory and experimental
approaches of AFM. The section begins with a description
of the imaging AFM as well as its different modes of
operation, which allow for its applications in various
experimental protocols. Later, force spectroscopy is
described. Focus is placed on the force apparatus used
in our laboratory, which relies on the same basic principals
as the commercially available AFMs. In addition to a
description of its operation, this section also discusses
the different applications of the force apparatus.
Imaging AFM
Optical Beam Deflection. An AFM has a force sensor
called a cantilever to measure the forces between a sharp
tip and the surface (23). Unlike the optical microscope that
relies on 2D images, the images acquired with the AFM are
obtained in three dimensions: the horizontal xy-plane and
the vertical z-direction. As shown in Fig. 4, the tip at the
end of the cantilever is brought in close proximity to the
sample mounted on a piezoelectric element. The AFM can
be compared with a record player such as an old stylusbased instrument (1). It combines the principles of a scanning tunneling microscope (STM) and the stylus profiler.
However, the probe forces in the AFM are much smaller
than those ( 104 N) achieved with a stylus profiler.
Laser
Photodiode
Preamp
A/D
Converter
Bimorph
actuator
Sample
Cantilever & Tip

Computer & Interface
SPM controller
Piezoelectric
tube scanner
X, Y, Z signals
Figure 4. Schematic illustration of an AFM. A force sensor

consists of a flexible cantilever with an extremely sharp tip at
its end. A ceramic (Si3N4) or semiconductor (Si) tip on the
cantilever can be brought into close proximity to a sample
surface. As the tip is close to the sample surface, it either
continuously touches or periodically vibrates on the surface, and
bends or changes in its vibration amplitude and frequency. A laser
spot is reflected off the top of the cantilever. When the cantilever
bends, the laser light is deflected onto a two-panel photodiode. The
detected signals are amplified and transformed by electronic
circuits and sent to an SPM controller. The SPM controller, the
computer, and their interfaces generate an image.
The AFM belongs to the family of scanning probe microscopes (SPMs). Like all other SPMs, the AFM uses an
extremely sharp tip that moves over the sample surface
with a raster scan (like the movement of the electron beam
on the TV screen). In the first AFM, the bending of the
cantilever was detected using an STM, but now a sensitive
and simple optical method is used in most AFMs (24). As
shown in Fig. 4, a laser beam is reflected off the cantilever
onto a two-panel photodiode. As the cantilever bends, the
position of the reflected laser light changes. Measurements
are obtained as a result of the differences in the signal
between the two segments of this photo-detector.
Feedback Operation. The AFM uses a piezoelectric
element to position and scan the sample with high resolution. A tube-shaped piezoelectric ceramic that has a high
stability is used in most SPMs. Application of voltage
results in the stretching or bending of the piezoelectric
tube, allowing it to move in all three dimensions and to
position the cantilever probe with very high precision. For
example, by applying a voltage to one of the two electrodes
(xy-axis) the tube scanner expands and tilts away from a
center position (xy-origin). A corresponding negative voltage applied to the same electrode causes the tube scanner
contract, resulting in movements on the xy-plane relative to
the origin. The magnitude of the movement depends on the
type of piezoelectric ceramic, the shape of the element, and
the applied voltage.
Feedback control is used for many common applications,
such as thermostats, which are used to maintain a particular temperature in buildings, and autopilot, commonly
used in airplanes. In the AFM, a feedback loop is used to
keep the force acting on the tip in a fixed relationship with

the surface while a scan is performed. The feedback loop
consists of the piezoelectric tube scanner, the cantilever
and tip, the sample, and the feedback circuit. The feedback
circuit consists of proportional and integral gain controls
and provides an immediate response to scanning parameter changes. A computer program acts as a compensation network that monitors the cantilever deflection and
attempts to keep it at a constant level.
Contact Mode (Static Mode). The AFM operates by

measuring the intermolecular forces between the tip and
sample. The most common method used in imaging AFM is
contact mode, where the piezoelectric element slightly
touches the tip to the sample. The experimental setup is
shown in Fig. 4. As a result of the close contact, the tip and
sample remain in the repulsive regime of the tip-sample
interaction shown in Fig. 5. Thus, the AFM measures
repulsive force between the tip and sample. As the raster
scan moves the tip along the sample, the two-panel photodiode measures the vertical deflection of the cantilever,
which reveals the local sample height. Each contour of the
Force
Contact mode
Repulsive
region
Intermittent contact
Tip-sample distance
Attractive
region
Non-contact
Figure 5. Relationship between the operating modes of AFM and

the force regions. The x-axis of the graph is the tip-sample distance
and the y-axis is the force or potential. The graph shows the force or
potential, as a function of the distance, simply calculated by the
LennardJones potential and the DMT approximation. In contact
mode AFM, the tip-sample interaction lies in the repulsive region
of the curve above the x-axis. As the cantilever is pushed upward, a
resulting restoring force occurs, which can be given by Hookes
Law. The difference between this graph and the measured forcedistance curve occurs when the force measurement is not static but
dynamic and is quickly affected by the capillary force in the thin
surface water layer. In intermittent contact mode, the cantilever is
operated by a vibration of an amplitude of 10 to 100 nm. The tip is
still in contact with the surface, but it just contacts or taps on the
surface for a very small fraction of its oscillation period. In this
operation mode, the tip-sample interaction is broad, ranging from
the repulsive region of the curve to the attractive region due to the
long-range van der Waals force. The tip-sample interaction in
noncontact mode is much weaker than the one in contact and
intermittent contact mode. The force between the tip and sample is
several orders of magnitude weaker than in the contact regime. As
the cantilever in noncontact mode is vibrated near its resonance
frequency with an amplitude less than 10 nm, the spacing between
the tip and sample is on the order of several to tens of nanometers.
507
surface results in a movement of the tip in the xyz-direction,

resulting in a change in the deflection angle of the laser
beam. This change is measured through the photodiode
and translated finally to an image.
Tip-sample Interaction: The cantilever in the AFM is a
critical component (1). The force produced by a spring
always tends to restore the spring to its equilibrium position. When the spring is pushed upward by a distance z, it
has to be pulled downward. This restoring force is given by
Hookes Law as:
Fz k z z0
()
where k is a spring constant and depends on the material

and dimensions of the cantilever, z is the vertical position of
the cantilever, and z0 is the equilibrium position. As a
cantilever with a spring constant of 0.1 newton/meter (N/
m) is moved by 1 nm, the resulting force is 0.1 nanonewton
(nN). The first tip used by the inventors of the AFM was
made from diamond glued to a lever of gold foil (1). Microfabricated cantilever tips are now commercially used.
Electromagnetic forces determine the properties of
solids, liquids, and gases; the behavior of particles in
solution; and the organization of biological structures
(25). These forces are also the source of all intermolecular
interactions including covalent bonds, Coulomb forces,
ionic forces, ion-dipole interaction, and dipole-dipole interaction. In the AFM, the intermolecular interactions
between the tip and the sample surface include van der
Waals forces, electrostatic forces, water capillary force, and
material properties including elasticity. The most common
force in the tip-sample interaction is the van der Waals
force. The force is calculated using the LennardJones
potential, which combines the attractive van der
Waals and the repulsive atomic potentials (25). The force
depends on the distance between the tip and the sample, as
shown in Fig. 5. This calculated force is an estimate of the
van der Waals forces and is usually a few nanonewtons in
magnitude.
Force-distance Curve: Since the invention of the AFM,
many researchers have used it to measure the tip-sample
interaction force on the atomic scale. The AFM records the
force as the tip is brought in close proximity to the sample
surface, even indented into the surface, and then pulled off.
The measured force curve is a plot of cantilever deflection
versus the extension of the z-piezoelectric scanner
(z-piezo). A force-distance curve is a kind of interpretation
of the force measurements. It needs a simple relationship
between the cantilever deflection and the tip-sample distance. Thus, the force-distance curve describes the tipsample interaction force as a function of the tip-sample
distance rather than as a function of the z-piezo position. It
is difficult to measure the quantitative forces with this
technique because the spring constant of the cantilever and
the shape of the tip are not accurately known. However,
this technique has been used to study adhesion, elasticity,
bond rupture length, and even thickness of adsorbed
layers. These studies of the fundamental interactions
between the sample surfaces have extended across basic
508
science, chemistry, biology, and even material science. The

interaction force between tip and sample is typically on the
order of tens of pN for biomolecular interactions. Force
measurements in solution have the advantages of the AFM
due to the lower tip-sample interaction.
Constant Force and Constant Height: In contact mode, the
tip is scanned across the surface in contact either at a
constant force or at a constant height above the sample.
Constant force mode is achieved by use of a z-feedback loop
from the deflection signal. The feedback circuits serve to
maintain a constant force between the tip and the sample
while the tip follows the contours of the surface. The
piezoelectric tube can respond to any changes in the cantilever deflection. A computer program acts to keep the
cantilever deflection at a constant level. Then, the tipsample interaction can be kept at a predetermined restoring force. This technique is used to observe the precise
topography of the sample surface. If the z-feedback loop is
switched off, then the z-direction of the piezoelectric tube is
kept constant, and an image is generated based on the
cantilever deflection. Using constant height can be useful
for imaging very flat samples.
Lateral Force Microscopy: Lateral force microscopy
(LFM) is an extension of contact mode, where an additional
detected parameter is the torsion of the cantilever, which
changes according to the friction force (26). This lateral
force induces a torsion of the cantilever, which, in turn,
causes the reflected laser beam to undergo a change in a
perpendicular direction to that resulting from the surface
corrugation. The LFM uses a photodiode with four segments to measure the torsion of the cantilever. When the
cantilever is scanned across the surface in contact, differences in friction between tip and sample cause the tip to
stick-slip on the surface. This stick-slip behavior creates a
characteristic saw-tooth waveform of atomic level in the
friction image (27). The LFM can provide material-sensitive contrast because different components of a composite
material exert different friction forces. Researchers often
call this operation mode friction force microscopy (27,28).
Increasing wear with decreasing sliding velocity on the
nanometer scale has been observed with this technique. It
has been demonstrated with LFM that, on the atomic scale,
frictional properties are sensitive to changes in surface
properties on chemical modification. The LFM can also be
applied to chemical force microscopy (CFM) by a modified
tip with chemical functionality (29). It has been demonstrated with CFM that mapping the spatial arrangement of
chemical functional groups and their interactions is of
significant importance to problems ranging from lubrication and adhesion to recognition in biological systems.
Capillary Force: The thin surface water layer that exists
on the sample surface will form a small capillary bridge
between the tip and the sample. The capillary force is
important when the AFM is operated in air. Examine
the effect of surface tension on AFM measurements. At
the moment of tip contact with a liquid film on a flat
surface, the film surface reshapes producing a ring around
the tip. The water layer wets the tip surface because the
water-tip contact (if it is hydrophilic) is energetically

advantageous as compared with the water-air contact. If
the tip radius is 10 nm and the contact angle is small (i.e.,
hydrophilic), a capillary force of about 10 nN can result.
Thus, the capillary force is the same order of magnitude as
the van der Waals interaction. An AFM tip has been used to
write alkanethiols with a 30 nm line-width resolution on a
gold thin film in a manner analogous to that of a dip pen
(30). Recently, this dip-pen nanolithography has also been
applied to direct nanoscale patterning of biological materials such as DNA, peptides, and proteins on glass substrates.
Vibration Mode (Dynamic Mode). In dynamic mode, the
cantilever is oscillated close to its resonance frequency.
This vibration mode operates at a frequency-modulation
(FM) mode or the more common amplitude-modulation
(AM) mode, which are basically the same as the frequencies
used in radio communication. In the FM mode, a z-feedback
loop keeps a constant force between the tip and the sample
while the tip follows the contours of the surface by maintaining the resonance frequency. In the AM mode, the zfeedback loop keeps the constant tip-sample interaction by
maintaining the amplitude of oscillation.
Intermittent Contact Mode: The cantilever in dynamic
mode can easily be vibrated by a piezoelectric ceramic
called a bimorph actuator. In air, the cantilever is oscillated close to its resonance frequency and positioned above
a sample surface. When the vibrating cantilever comes
close to the surface, its oscillation amplitude may change
and can be used as the control signal. In this AM mode, the
tip is still in contact with the surface, but it just contacts or
taps on the surface for a very small fraction of its oscillation period. This operation mode is best known as tapping
mode in commercial AFMs and, more generally, as intermittent contact mode.
As a raster scan moves the tip on the sample, the foursegment photodiode measures the vibration signal of the
cantilever. The detected signal can be changed to root meansquare values by an analog-to-digital converter. In constant
force mode, the z-feedback loop adjusts so that the averaged
amplitude of the cantilever remains nearly constant. Each
contour of the surface causes a movement of the tip in the
xyz-direction, resulting in a change of the oscillation amplitude of the cantilever. This change is measured through a
photodiode and finally translated to an image. In air, friction
forces due to the surface water layer are dramatically
reduced as the tip scans over the surface. Tapping mode
may be a far better choice than contact mode for imaging of
biological structures due to their inherent softness. In tapping mode, the cantilever can be vibrated at an amplitude of
less than 100 nm. Additionally, changes in the phase of
oscillation under tapping mode can be used to discriminate
between different types of materials on the surface.
Tip-Sample Interaction: The mechanical resonance of
the cantilever plays a major role in the response of the
system for an interaction between a tip mounted on a
vibrating cantilever and a non-homogeneous external force
(23). Although basic equations governing the operation of a
bimorph actuator used to vibrate the cantilever are not

introduced here, the position of the bimorph is given by:
u u0 Aex cosvt
where u0 is the equilibrium position and the excitation is
done with amplitude Aex, a frequency v, and a phase shift
f. The fundamental resonance frequency of the cantilever
can be approximately calculated from equating its strain
energy at the maximum deflection to the kinetic energy at
the point of zero deformation. A more accurate method,
which takes into consideration all the resonance frequencies of the cantilever together with the modes of vibration,
can be obtained by solving the equation of motion subject to
the boundary conditions (23). A basic equation to describe
the motion of the cantilever is briefly introduced. If the tipsample interaction is uniform and includes dissipative
force in Newtons second law, the vibration system including the cantilever can be described as follows:
Fz kz u ydz=dt md2 z=dt2
()
where F(z) is the tip-sample interaction force, k is a spring

constant of the cantilever, z is the vertical position of the
cantilever, u is the motion of the bimorph, g is the dissipation term (i.e., the friction coefficient of the material or the
environment), and m is the effective mass of the cantilever.
For the constant amplitude mode, we assume that the
frictional force g (dz/dt) is compensated for by the driving
force Fex k Aex cos(vt f). Then, the equation of motion is
reduced to F(z) k z m(d2z/dt2). If a strong tip-sample
interaction occurs only at the point of contact, the motion
of the cantilever tip can be almost perfect harmonic oscillation, z zo A sin vt.
Resolution and Tip Effects: The resolution obtained by
an AFM depends greatly on the sharpness of the cantilever
tip. Broadening effects usually develop when imaging biological structures having extremely small surface features
like a DNA strand (4). If a tip with a radius of curvature of
about 20 nm is used to image DNA on a substrate surface,
the observed width is about 20 nm, which is considerably
greater than the expected value of 2.4 nm deduced from the
van der Waals radii of DNA. When the tip radius is
comparable with the size of the feature being imaged, it
is important to evaluate the radius of the tip end. As such,
the development of sharper tips is currently a major concern for commercial vendors, which is also of interest for
biologists whose work would greatly benefit from much
faster scanning. Recently, improvement of the scanning
speed in AFM is one of the most important topics. The tipsample interaction also tends to distort biological structures because they are relatively soft (31).
Phase Imaging: Phase imaging is an extension of tapping mode based on the measurement of the cantilever
phase lag (32). The dependence of phase angles in tapping
mode AFM on the magnitude of tip-sample interactions has
been demonstrated. The phase images of several hard and
soft samples have been recorded as a function of the free
amplitude and the reference of the tapping amplitude. It is
thought that the elastic deformation associated with the
509
tip-sample repulsive force can be estimated by the repulsive contact interaction. In many cases, phase imaging
complements the LFM and force modulation techniques,
often providing additional information along with a topographic image. Phase imaging like LFM can also be applied
to CFM by using a modified tip with chemical functionality.
Pulsed Force Mode: Pulsed force mode (PFM) is a nonresonant and intermittent contact mode used in AFM
imaging (33). It is similar to tapping mode in that the
lateral shear forces between the tip and the sample are
also reduced. In contrast to tapping mode, the maximum
force applied to the sample surface can be controlled, and it
is possible to measure more defined surface properties
together with topography. This mode is similar to the force
modulation techniques of CFM in that a chemically modified tip is used. A series of pseudo force-distance curves
can be achieved at a normal scanning speed and with much
lower expenditure in data storage. A differential signal can
be amplified for imaging of charged surfaces in terms of an
electrical double-layer force.
Noncontact Mode: A reconstructed silicon surface has
been imaged in a noncontact mode by AFM with true
atomic resolution (34). The operation of the AFM is based
on bringing the tip in close proximity to the surface and
scanning while controlling the tip-sample distance for the
constant interaction force. The tip-sample interaction
forces in noncontact mode are much weaker than those
in contact mode, as shown in Fig. 5. The cantilever must be
oscillated above the sample surface at such a distance as is
included in the attractive regime of the intermolecular
force. Most surfaces are covered with a layer of water,
hydrocarbons, or other contaminants when exposed to
air, which makes it very difficult to operate in ambient
conditions with noncontact mode. Under ultrahigh
vacuum, clean surfaces tend to stick together, especially
when the materials are identical. The FM mode used in
noncontact mode can keep the constant tip-sample interaction by maintaining the resonance frequency of oscillation through the z-feedback loop. Nearly ten years
following the invention of the AFM, a few groups achieved
true atomic resolution with a noncontact mode (35). After
that, several groups succeeded in obtaining true atomiclevel resolution with noncontact mode on various surfaces. Many important yet unresolved problems, such as
determining the tip-sample distance where atomic-level
resolution can be achieved, still remain. Experimentally,
atomic-level resolution can be achieved only between 0 and
0.4 nm. A stiff cantilever vibrates near resonance frequency (300400 kHz) with amplitude of less than 10 nm.
In covalently bound materials, the charge distribution
of surface atoms reflects their bonding to neighboring
atoms (36). These charge distributions have been imaged
by noncontact mode with a light-atom probe such as a
graphite atom. This process revealed features with a lateral distance of only 77 picometers (pm). However, all of the
atomic-scale images have been generated in ultrahigh
vacuum, which has limited applications in biology.
Recently, several groups have reported obtaining
atomic-scale images with FM mode in ambient conditions
510
and liquid environments. In the near future, true atomiclevel imaging by AFM will be commercially available in
various environments.
Force Spectroscopy
Equipment
AFM Instrumentation. The AFM that is used in the
authors laboratory is a homemade modification of the
standard AFM design that is used for imaging and is shown
in Fig. 5 (19). It operates on the same basic principles as a
commercial AFM. In the authors design, improvement of
the signal quality by reducing mechanical and electrical
noise and improvement of the instruments sensitivity by
uncoupling the mechanisms for lateral and vertical scans
was achieved. The cantilever is moved vertically up and
down using a piezoelectric translator (Physik Instrumente,
model P-821.10) that expands or contracts in response to
applied voltage. The vertical range of the piezo is 015 mm.
A dish coated with substrate is placed below the cantilever,
and the cantilever with a cell or protein attached can be
lowered onto that dish using the piezo allowing for the
receptor-ligand interaction to take place. During the acquisition of a force scan, the cantilever is bent (Fig. 6) causing
the beam of a 3 mW diode laser (Oz Optics; em. 680 nm)
that is focused on top of the cantilever to be deflected. A
two-segment photodiode (UDT Sensors; model SPOT-2D)
monitors these deflections of the laser beam. An 18 bit
optically isolated analog-to-digital converter (Instrutech
Corp., Port Washington, NY) then digitizes the signal from
the photodiode. Custom software is used to control the
piezoelectric translator and to time the measurements.
The AFM is shielded inside of an acoustic/vibration isolation chamber in order to reduce vibration and aid in keeping a stable temperature. The detection limit of the AFM
system is in the range of 20 pN.
Cantilever Calibration. It is necessary to determine the
spring constant of the cantilever kC (i.e., F kCx) in order to
translate the deflection of the cantilever x to units of force
F. Calibrating the cantilever can be achieved through
theoretical techniques that provide an approximation of
kC (37) or through empirical methods. Using empirical
methods to determine kC involves taking measurements
of cantilever deflection with application of a constant
known force (38) or measuring the cantilevers resonant
frequency (39). The method the authors use for calibrating
cantilevers is based on Hutter and Bechhoefer (39). The
Figure 6. AFM experimental set-up. (a) Schematic diagram of the

AFM. (b) Photograph of the authors AFM setup. The CCD camera
is not in view.
authors use triangleshaped unsharpened gold-coated silicon-nitride cantilever tips that have spring constants ranging from 10 mN/m to 50 mN/m for ligand-receptor force
measurements (TM Microscopes, Sunnyvale, CA). The
cantilever tip can be treated as a simple harmonic oscillator
whose power spectrum of thermal fluctuation can be used
to derive the spring constant, which can be achieved by
raising the cantilever a few microns from the surface of the
experimental dish and monitoring its natural vibrational
frequency for 23 s. Each vibration mode of the cantilever
receives the thermal energy commensurated to one degree
of freedom, kB T/2. The measured variance of the deflection
hxi2, can then be used to calculate the spring constant (i.e.,
kBT kChxi2, where kB and T are Boltzmanns constant and
temperature, respectively). To separate deflections belonging to the basic (and predominant) mode of vibration from
other deflections or noise in the recording system, the
power spectral density of the temperature-induced deflection is determined. The spring constant is estimated using
only the spectral component corresponding to the basal
mode of vibration. The spring constant can be calibrated in
either air or solution using this approach. The calculated
spring constant kC can then be used to calculate rupture
force F by F kCCD V. DV is the change in voltage detected
by the photodiode just prior to and immediately after the
rupture event. C is a calibration constant that relates
deflection and photodiode voltage and is determined
from the deflection of the cantilever when it is pressed
against a rigid surface, such as the bottom of a plastic petri
dish (19).
Applications
Receptor-Ligand Adhesion Measurements. Bell Model:
AFM force measurements (Fig. 7) of ligand-receptor interactions can be used to determine the dynamic strength of a
complex and characterize the changes in free energy that
the particular complex undergoes (i.e., energy landscape)
during its breakage. The Bell model can be used to interpret these measurements (40). The Bell model is based on
the assumption that the application of an external mechanical force to a receptor-ligand interaction bond will reduce
the activation energy that needs to be overcome in order to
break this bond. This unbinding force should increase with
the logarithm of the rate at which an external mechanical
force is applied toward the unbinding of adhesion
complexes (i.e., loading rate), which was confirmed by a
number of studies. For example, studies using the biomembrane force probe (BFP) (40) and the AFM have shown that
increases in loading rate cause an increase in rupture force
between individual complexes of streptavidin/biotin
(12,15,41).
AFM Measurements of Adhesive Forces: In order to carry
out force measurements, a cell is first attached to a cantilever tip and another cell or substrate proteins are plated
on a dish. The method employed to attach cells to the
cantilever tip works best on nonadherent items. A cell is
attached to the AFM cantilever via concanavalin A (con A)mediated linkages (15). Most cells have receptors for con A
on their surface and will attach to the tip. To prepare the
con A-functionalized cantilever, the cantilevers are soaked
511
AFM cantilever
Approach
3A9
Immobilized ICAM-1
Contact
Retract
Separation
Figure 7. Steps in the acquisition of an AFM force measurement.
The first step is the approach of the cantilever with a cell bound to
the substrate, followed by contact between the cell and substrate
and retraction of the cantilever, which results in the separation of
the cell from the substrate. The cantilever is bent during this
process. The arrows indicate the direction of cantilever movement.
in acetone for 5 min, UV irradiated for 30 min, and incubated in biotinamidocaproyl-labeled bovine serum albumin
(biotin-BSA, 0.5 mg/ml in 100 mM NaHCO3, pH 8.6; Sigma)
overnight. The cantilevers are then rinsed three times with
phosphate-buffered saline (PBS, 10 mM PO43, 150 mM
NaCl, pH 7.3) and incubated in streptavidin (0.5 mg/ml
in PBS; Pierce, Rockford, IL) for 10 min at room temperature. Following the removal of unbound streptavidin, the
cantilevers are incubated in biotinylated Con A (0.2 mg/ml
in PBS; Sigma) and then rinsed with PBS.
The actual process of attaching a cell to a cantilever tip
is reminiscent of fishing. A cantilever tip is positioned
above the center of the cell. The largest triangular cantilever (320 mm long and 22 mm wide) with a spring constant
of 0.017 N/m on the cantilever chip is usually used in our
measurements. The cell is brought into focus, with the
cantilever slightly out of focus. Then, the tip is lowered
onto the center of the cell and held there motionless for
approximately 1 s. When attached, the cell is positioned
right behind the AFM cantilever tip. The force required to
dislodge the cell from the tip is greater than 2 nN, which is
much greater than the forces measured in the receptorligand studies that were, at most, 300 pN (15).
A piezoelectric translator is used during measurement
acquisition to lower the cantilever with an attached cell onto
the sample. The interaction between the attached cell and
the sample is given by the deflection of the cantilever. This
deflection is measured by reflecting a laser beam off the
cantilever into a position sensitive two-segment photodiode
detector, as described in the instrumentation section above.
Figure 8. AFM force versus displacement traces of the

interaction between cells expressing LFA-1 and immobilized
ICAM-1. (a) Multiple-bond measurements acquired with a
compression force of 200 pN, 5 s contact, and a cantilever
retraction speed of 2 mm/s. The measurements were carried out
with a resting, untreated cell (1st trace), a Mg2-treated cell (2nd
trace), and a PMA-stimulated cell (3rd trace). The 4th trace
corresponds to a measurement acquired from a PMA-stimulated
cell in the presence of LFA-1 (20 mg/ml FD441.8) and ICAM-1
(20 mg/ml BE29G1) function-blocking monoclonal antibodies
(mAbs). Arrows point to breakage of LFA-1/ICAM-1 bond(s). fde
is the detachment force, and the shaded area estimates the work of
de-adhesion. (b) Single-molecule measurements of LFA-1/ICAM-1
unbinding forces. Traces 2 and 5 show adhesion. Measurements
were obtained under conditions that minimized contact between
the LFA-1-expressing cell and the ICAM-1-coated surface. The
compression force was reduced to 60 pN and the contact time to
50 ms. An adhesion frequency of less than 30% in the force
measurements ensured that a > 85% probability exists that the
adhesion event is mediated by a single LFA-1/ICAM-1 complex
(42,43). The frequency of adhesion in test and control experiments
was examined to confirm the specificity of the interaction. The
addition of monoclonal antibodies against either LFA-1 or ICAM-1
significantly lowered the frequency of adhesion of both resting cells
and activated cells under identical experimental conditions.
As a result of this process, a force scan is obtained. The

studies shown in Fig. 8 (42,43) were conducted on cells
expressing the adhesion receptor LFA-1 (leukocyte function-associated antigen-1), an integrin expressed on the
surface of T-cells, and its ligand ICAM-1 (intercellular
adhesion molecule-1), expressed on the surface of APCs.
In these experiments, LFA-1-expressing cells and ICAM-1
protein were used. An example of a few force scans from
multiple bond cell adhesion studies can be seen in Fig. 8a.
The red trace is the approach trace and the black is the
retract trace. As the cantilever is lowered and contact is
made between the cell and substrate, an initial increase in
force occurs. As the cantilever is retracted back up, the
force returns to zero and becomes negative as bonds are
stretched and begin to break. The jumps in the force scan,
which are pointed out by the arrows, represent bonds
breaking. Two parameters can be used in such measurements to assess the level of cell adhesion. One is the
detachment force, which is the maximum force required
to dislodge the cell. Another is the work of deadhesion,
which is the amount of work required to pull and stretch
512
the cell and for the bonds to break. It is derived by integrating the adhesive force over the distance traveled by the
cantilever. In this example, Mg2 and PMA are used, which
enhance the adhesion of the cells studied through various
mechanisms. It is easily observed that a very pronounced
increase occurs in the area under the curve as well as the
number of bonds that were broken following the application of these adhesion stimulators (16).
AFM tip
3A9 cell
PMA
1.0
B
Indentation (m)
FM Force Measurements of Individual Receptor/Ligand

Complexes: A different set of information can be derived
from single-molecule adhesion measurements. These type
of studies offer insight into the dissociation pathway of a
receptor-ligand interaction and the changes in free energy
that are associated with this process, which is achieved by
measuring single-bond interactions between a receptor
and ligand at increasing loading rates (20 pN/s
50,000 pN/s). In the authors setup, it translates to using
rates of retraction of the cantilever from 0.1 to 15 mm/s.
In order to obtain unbinding forces between a single
receptor-ligand pair, the experiments have to be carried
out in conditions that minimize contact between the cantilever tip and substrate. A > 85% probability exists that the
adhesion event is mediated by a single bond if the frequency of adhesion is maintained below 30% (15,42). An
example of such measurements can be seen in Fig. 8b.
Depending on the speed at which the cantilever is
retracted during the measurements, the collected data
usually needs to be corrected for hydrodynamic drag,
which is due to the hydrodynamic force that acts in the
opposite direction of cantilever movement, and its magnitude is proportional to the cantilever movement speeds.
The hydrodynamic force may be determined based on the
method used by Tees et al. and Evans et al. (42,43). In
single-bond AFM studies, it is found that the data
obtained with cantilever retraction speeds higher than
1 mm/s needed to be corrected by adding the hydrodynamic
force.
The damping coefficient can be determined by plotting
the hydrodynamic force versus the speed of cantilever
movement. The damping coefficient is the slope of the
linear fit and was found to be about 2 pN &s/mm in the
authors work (15).
Resting
0.5
Mg2+/EGTA
0
0
200
300
Force (pN)
Figure 9. Acquisition of cell compliance measurements. (a) Tip of

the AFM cantilever indenting a 3A9 cell. The cell compliance
measurements were based on the assumption that the cell is an
isotropic elastic solid and the AFM tip is a rigid cone (4446).
According to this model, initially proposed by Love and Hertz, the
force (F)-indentation (a) relation (shown) is a function of Youngs
modulus of the cell, K, and the angle formed by the indenter and
the plane of the surface, u, as in equation (1). The indenter angle, u,
is assumed formed by the AFM tip and the 3A9 cell to be 558 and
the Poisson ratio n to be 0.5. (b) Force versus indentation traces of
resting, PMA-stimulated and Mg2-treated 3A9 cells.
solid and the AFM tip is a rigid cone (4446). According

to this model, initially proposed by Hertz, the force (F)
indentation (a) relation is a function of Youngs modulus of
the cell, K, and the angle formed by the indenter and the
plane of the surface, u, as follows:
F
AFM Measurements of Cell Elasticity. The AFM can also

be used as a microindenter that probes the mechanical
properties of the cell. In these measurements, which enable
assesment of cell elasticity, a bare AFM tip is lowered onto
the center of the cell surface at a set rate, typically 2 mm/s.
Following contact, the AFM tip exerts a force against the
cell that is proportional to the deflection of the cantilever.
The indentation force used is below 1 nN ( 600 pN) in
order to prevent damage to the cell. The deflection of the
cantilever is recorded as a function of the piezoelectric
translator position during the approach and withdrawal
of the AFM tip. The force-indentation curves of the cells are
derived from these records using the surface of the tissue
culture dish to calibrate the deflection of the cantilever.
Then, one can estimate the Youngs modulus, which is a
measure of elasticity. Estimates of Youngs modulus are
made on the assumptions that the cell is an isotropic elastic
100
K
4
a2
21 v2 p tan u
Youngs modulus are obtained in the authors laboratory by

least square analysis of the forceindentation curve using
routines in the Igor Pro (WaveMetrics, Inc., Lake Oswego,
OR) software package. The indenter angle u and Poisson
ratio n are assumed to be 558 and 0.5, respectively.
In order to determine the cells elasticity, the force
versus indentation measurements are fitted to the curves
of the Hertz model. Figure 9 (4446) illustrates an example
of such measurements acquired on cells of varying elasticity. Cells with the greatest degree of indentation at a
particular applied force will have the lowest Youngs modulus values and will therefore be the softest.
Protein Folding/Unfolding. The AFM can also be used to
study protein unfolding. A cantilever tip is used to pick up
proteins attached to a surface, which is followed by retrac-
Figure 10. Consecutive unfolding peaks of a titin polyprotein,

composed of FNIII domains. The inset demonstrates the
corresponding steps of the unfolding process in correlation to
the AFM data.
tion of the cantilever, which results in protein unfolding.

The length of the unfolded protein can be over ten times its
folded length, depending on the protein being studied (9).
This forced protein unfolding generates an opposing
force due to the sudden drop in entropy as the protein is
unfolded. Although a lower force is required to begin
unfolding the protein, the force required to continue the
unfolding is increased rapidly as the protein approaches its
full, unfolded length. This phenomenon has been described
by the worm-like chain model (WLC) of elasticity. The WLC
model is based on two parameters, the total or contour
length of the polymer being stretched and the persistence
length. The persistence length reflects the polymer flexibility and is the length attained when a polymer is bent. A
smaller persistence length is an indication of higher
entropy and a polymer that is more difficult to unfold
(47). When a multidomain protein is extended using the
AFM, the first domain is unfolded at a certain pulling force,
followed by a return of the cantilever to zero. Further
unfolding meets with resistance once again, resulting in
a characteristic saw-tooth profile of the unfolding, with
each domain that was unfolded being represented by a
peak. Figure 10 from Andreas F. Oberhauser illustrates
this process. It is the unfolding of a titin polyprotein, which
is composed of FNIII domains (22). The protein can also be
refolded following unfolding, which is done by bringing the
cantilever back down to the substrate and once again
retracting it. If force curves representative of unfolding
are observed once again, then refolding most likely took
place. It is a much slower process (on the order of seconds)
than the forced unfolding (48).
EVALUATION
Imaging AFM
The AFM is an exciting novel technology that enables the
study of biological structures under physiological conditions. The AFM is probably the only technique of its kind
513
that enables image dynamic processes taking place in real

time. A number of other techniques are currently available in the biological sciences for imaging studies, however, most result in modifications to the biological sample.
One such technique is electron microscopy (EM), which,
until recently, provided images of the highest resolutions.
In recent years, a number of modifications to the AFM
have brought the resolution up to par and even surpassed
those of EM.
In recent years, many advances have been made in the field
of AFM. Significant improvements in resolution have been
gained through cantilever tip modification. The currently
available cantilevers are relatively soft and flexible with
spring constants of 0.010.5 N/m. Tip deformation is one
aspect that limits resolution. Recently, stiffer cantilevers have
been designed improving resolution. One example are quartz
cantilevers with spring constants on the order of 1 kN/m
allowing for possibly subatomic-level resolution (34). Smaller
cantilevers have been designed that increase the possible
scanning speed of the AFM. Images of 100
100 pixels
(240 nm scan size) have been achieved in 80 ms. A sharper,
finer tip can also improve resolution, which has been achieved
through the use of carbonanotubes, probably the greatest
probe improvement to date, which are seamless cylinders
composed of sp2-bonded Carbon (49).
Several characteristics exist that make Carbon nanotubes improved AFM tip materials, including small diameter, a high aspect ratio, large Youngs modulus, and
mechanical robustness. They are able to elastically buckle
under large loads. All these properties translate into
higher sample resolution. Chemical vapor deposition has
made it easier to grow Carbon nanaotubes on cantilever
surfaces, a process that replaces previously more laborious
and time-consuming attachment techniques (4,50).
A few techniques also exist worth mentioning that have
improved AFM imaging. One such method is cryoAFM.
This method addresses the previously mentioned problem
of tip and sample flexibility. In this case, samples are
imaged at extremely cold temperatures in their cryogenic
states, which provides a rigid surface that exhibits a high
Youngs modulus, thus reducing cantilever deformation
and increasing resolution. CryoAFM images match and
surpass EM images. Other improvements address the
problem resulting from the vibration induced by the commonly used piezoelectric translator. This vibration is translated to the cantilever holder and the liquid containing the
sample being imaged. Magnetic mode (MAC) eliminates
the cantilever holder entirely and replaces it with a magnetic cantilever. The cantilever is manipulated via a magnetic field. Photothermal mode (PMOD) uses a bimetallic
cantilever that is oscillated via a pulsed diode laser (50,51).
Advances have also been made in single-molecule
manipulation with a nanomanipulator. This method relies
on a force feedback pen that actually allows the user to
touch and manipulate the sample being studied. For example, one can dissect DNA from a chromosome. The interaction forces involved during manipulation of samples can
also be studied through oscillating mode imaging. The
nanomanipulator can now measure forces in the pNmN
range. For excellent reviews on this technique, see Yang
et al. (50) Fotiadis et al. (52).
514
Progress has also been made in imaging of membrane

proteins, which are not ideal candidates for X-ray crystallography, as they do not readily form 3D crystals. Atomiclevel resolution images have been obtained of membrane
proteins complexed with lipids using EM. However, AFM
images of these proteins offer an improvement in that they
can be carried out in near physiological conditions and
allow for the acquisition of functional and structural information (50,52).
The continued improvements leading to the enhanced
imaging capabilities of AFM are reflected in the most
recent work being done in the field. We would like to
highlight one area where a great deal of progress has been
made, which is the filed of DNA and RNA assembly of
nanostructures in which the imaging AFM plays a pivotal
role. Some of the earlier successes in this area included the
construction of 2D DNA arrays, which were assembled in a
predictable manner (53). Much progress has also been
made with RNA in an attempt to design self-assembling
building blocks. The goal of such studies is to generate
molecular materials, the geometry of which can be determined for applications in nanobiotechnology (5456).
Chworos et al. were able to form stable RNA structures
termed tectosquares from RNA helices without the presence of proteins (5). TectoRNAs can be thought of as
RNA Lego blocks that can be used for the formation of
supramolecular structures. In order for these structures to
be assembled, the right conditions have to be met in terms
of divalent ion concentrations, temperature, as well as the
strength, length, and orientation of the RNA. The AFM is
an essential tool used in this process as it allows the
researcher to obtain detailed images of the assembled
tectosquares providing the opportunity to compare predicted structures with those that actually formed. The
structures formed by Chworos et al. were in good agreement with the predicted structures. Figure 11 demon-
Figure 11. Diagram and AFM images of tectosquare

nanopatterns generated from 22 tectosquares. One micrometer
square scale AFM images obtained in solution (clockwise from the
upper leftmost image) for the ladder pattern, fish net pattern,
striped velvet pattern, basket weave pattern, cross pattern, tartan
pattern, polka dot pattern, lace pattern, and diamond pattern.
Scale bar, 20 nm.
strates the amazing predictability of assembly of these

structures, where nine different types of RNA patterns
were created (5). Such DNA and RNA structures may
have applications in nanotechnology and the material
sciences as they could be used to generate nanochips,
nanocircuits, and nanocrystals (57). For an excellent recent
review of DNA nanomechanical devices, please see
Seeman (58).
Force Spectroscopy
Force spectroscopy allows us to measure interaction forces
between receptors and their respective ligands. These
studies can be carried out with purified proteins or cells,
or a combination of both. Traditionally, adhesion measurements have been conducted using adhesion assays, which
involve the attachment of cells to dishes coated with substrate. The cells are later dislodged manually or using a
centrifuge, which are older yet still viable techniques that
provide basic kinetic information about the interaction of a
receptor-ligand pair. More advanced techniques for conducting force measurements include the use of microneedles, optical tweezers, magnetic beads, and the
biomembrane force probe. These techniques, much like
the AFM, provide more advanced information from which
energy landscapes of an interacting receptor-ligand pair
may be determined (11).
The AFM is also a powerful tool for determining the
mechanical properties of cells, which was traditionally
done using micropipettes or the cell poker, which offered
much less precision than the AFM. More recently, methods
such as the scanning acoustic microscope, optical tweezers,
and magnetic tweezers have also been used in addition to
the AFM (59).
An important advantage of the AFM over other methods
is that it can be used in conjunction with other techniques
through relatively simple modifications. Recently, it has
been combined with the patch clamp technique to study the
mechanically activated ion channels of sensory cells of the
inner ear. This strategy allowed the researchers to use the
AFM tip to stimulate the mechanosensory hair bundles by
exerting force on them and measure the electrical output of
the patch clamp simultaneously (9,60). Another example is
combining an AFM with a confocal microscope, which could
allow one to monitor cellular responses to AFM measurements using fluorescent reporter systems. One could monitor calcium levels, expression of caspases, and so on (61,62).
The AFM could also be combined with RICM microsopy as
well as FRET.
Other recent advances involve modifications that would
allow for more efficient and effective receptor-ligand studies, including the use of more than one cantilever on the
same chip simultaneously. In this case, multiple proteins
could be attached and their interaction with their ligand
could be measured. So far, this approach has been done
with two cantilevers, which involves the use of two laser
beams. Further modifications could allow for measurements with even more proteins. Improvements can also
be mad in plating of the ligands proteins. In the ideal
scenario, different proteins could be plated so that interaction between different receptor-ligand pairs cools be
carried out simultaneously. Current improvements also

involve finding better ways to attach cells and proteins
to the cantilever that would result in covalent attachment
to the tip. Another area that requires improvement is data
analysis. The currently available analysis program
involves days of rather tedious computer time. Automating
analysis would greatly reduce the time required to study a
particular interaction. Also, some improvements can be
made in data acquisition, where still frequent adjustments
of cantilever retraction speed and contact time are required
throughout the course of the experiment. Automating data
acquisition would allow experiments to be carried out
during the night, when noise levels are also minimal.
The applications of AFM technology are vast and too
numerous to describe in one review article. The authors
have attempted to summarize the technology that was
deemed to be of great importance in the developing field
of AFM. AFM technology is still limited to a relatively small
number of laboratories, which is most likely due to the lack
of familiarity with the field, limited expertise in operation,
as well as the expense involved in acquiring an AFM.
However, it is changing as more and more people are discovering the possibilities that become open to them if they
acquire and familiarize themselves with this technology.
BIBLIOGRAPHY
1. Binnig G, Quate CF, Gerber C. Atomic force microscope. Phys
Rev Lett 1986;56:930933.
2. Heinz WF, Hoh JH. Spatially resolved force spectroscopy of
biological surfaces using the atomic force microscope. Trends
Biotechnol 1999;17(4):143150.
3. Engel A, Lyubchenko Y, Muller D. Atomic force microscopy: A
powerful tool to observe biomolecules at work. Trends Cell
Biol 1999;9(2):7780.
4. Hansma HG. Surface biology of DNA by atomic force microscopy. Annu Rev Phys Chem 2001;52:7192.
5. Chworos A, et al. Building programmable jigsaw puzzles with
RNA. Science 2004;306(5704):20682072.
6. Kasas S, et al. Escherichia coli RNA polymerase activity
observed using atomic force microscopy. Biochemistry
1997; 36(3):461468.
7. Rivetti C, et al. Visualizing RNA extrusion and DNA wrapping
in transcription elongation complexes of bacterial and eukaryotic RNA polymerases. J Mol Biol 2003;326(5):14131426.
8. Pesen D, Hoh JH. Modes of remodeling in the cortical cytoskeleton of vascular endothelial cells. FEBS Lett 2005;579(2):
473476.
9. Horber JKH. Local probe techniques. Methods Cell Biol
2002;68:132.
10. Lee GU, Kidwell DA, Colton RJ. Sensing discrete streptavidinbiotin interactions with AFM. Langmuir 1994;10(2):354361.
11. Zhang X, Chen A, Wojcikiewicz E, Moy VT. Probing ligandreceptor interactions with atomic force microscopy. In:
Protein-Protein Interactions: A Molecular Cloning Manual.
Cold Spring Harbor, NY: Cold Spring Harbor Laboratory
Press; 2002. p 241254.
12. Yuan C, et al. Energy landscape of streptavidin-biotin complexes measured by atomic force microscopy. Biochemistry
2000;39(33):1021910223.
13. Benoit M, et al. Discrete interactions in cell adhesion measured by single molecule force spectroscopy. Nat Cell Biol
2000;2(6):313317.
515
14. Benoit M. Cell adhesion measured by force spectroscopy on

living cells. Methods Cell Biol 2002;68:91114.
15. Zhang X, Wojcikiewicz E, Moy VT. Force spectroscopy of
the leukocyte function-associated antigen-1/intercellular
adhesion molecule-1 interaction. Biophys J 2002;83(4):2270
2279.
16. Wojcikiewicz EP, et al. Contributions of molecular binding
events and cellular compliance to the modulation of leukocyte
adhesion. J Cell Sci 2003;116(12):25312539.
17. Matzke R, Jacobson K, Radmacher M. Direct, high-resolution
measurement of furrow stiffening during division of adherent
cells. Nat Cell Biol 2001;3(6):607610.
18. Hassan AE, et al. Relative microelastic mapping of living
cells by atomic force microscopy. Biophys J 1998;74(3):1564
1578.
19. Wojcikiewicz EP, Zhang X, Moy VT. Force and compliance
measurements on living cells using atomic force microscopy
(AFM). Biol Proced Online 2004;6:19.
20. Rief M, et al. Reversible unfolding of individual titin immunoglobulin domains by AFM [see comments]. Science
1997;276(5315):11091112.
21. Oesterhelt F, et al. Unfolding pathways of individual bacteriorhodopsins [see comments]. Science 2000;288(5463):143
146.
22. Li H, et al. Reverse engineering of the giant muscle protein
titin. Nature 2002;418(6901):9981002.
23. Sarid D. Scanning Force Microscopy. New York: Oxford
University Press; 1991.
24. Meyer G, Amer NM. Novel optical approach to AFM. Appl
Phys Lett 1988;53:10451047.
25. Israelachvili JN. Intermolecular and Surface Forces. 2nd ed.
London: Academic Press; 1992.
26. Meyer G, Amer NM. Simultaneous measurement of lateral
and normal forces with an optical-beam-deflection AFM. Appl
Phys Lett 1990;57(20):20892091.
27. Mate CM, et al. Atomic-scale friction of a tungsten tip on a
graphite surface. Phys Rev Lett 1987;59(17):19421945.
28. Overney RM, et al. Friction measurements on phasesegregated thin films with a modified atomic force microscope.
Nature 1992;359:133135.
29. Frisbie CD, et al. Functional group imaging by chemical force
microscopy. Science 1994;265:20712074.
30. Piner RD, et al. Dip-Pen nanolithography. Science 1999;
283(5402):661663.
31. Kwak KJ, et al. Topographic effects on adhesive force mapping of stretched DNA molecules by pulsed-force-mode
atomic force microscopy. Ultramicroscopy 2004;100(3
4):179186.
32. Magonov SN, EV, Whango MH. Phase imaging and stiffness
in tapping mode AFM. Surf Sci 1997;375:L385L391.
33. Miyatani T, Horii M, Rosa A, Fujihira M, Marti O. Mapping of
electrical double-layer force between tip and sample surfaces
in water with pulsed- forcemode atomic force microscopy.
Appl Phys Lett 1997;71:26322634.
34. Giessibl FJ, et al. Subatomic features on the silicon (111)(7
7) surface observed by atomic force microscopy. science
2000;289(5478):422426.
35. Morita SR, Wiesendanger R, Meyer E. Noncontact AFM. New
York: Springer; 2002.
36. Hembacher S, Giessibl FJ, MJ. Force microscopy with lightatom probes. Science 2004;305(5682):380383.
37. Sader JE. Parallel beam approximation for V-shaped atomic
force micrscope cantilevers. Rev Sci Instrum 1995;66:4583
4587.
38. Senden TJ, Ducker WA. Experimental determination of
spring constants in atomic force microscopy. Langmuir
1994;10: 10031004.
516
39. Hutter JL, Bechhoefer J. Calibration of atomic-force

microscope tips. Rev Sci Instrum 1993;64(7):1868
1873.
40. Bell GI. Models for the specific adhesion of cells to cells.
Science 1978;200:618627.
41. Merkel R, et al. Energy landscapes of receptor-ligand bonds
explored with dynamic force spectroscopy [see comments].
Nature 1999;397(6714):5053.
42. Tees DFJ, Woodward JT, Hammer DA. Reliability theory for
receptorligand bond dissociation. J Chem Phys 2001;114:
74837496.
43. Evans E. Probing the relation between forcelifetimeand
chemistry in single molecular bonds. Ann Rev Biophys Biomolec Struc 2001;30:105128.
44. Hoh JH, Schoenenberger CA. Surface morphology and
mechanical properties of MDCK monolayers by atomic force
microscopy. J Cell Sci 1994;107:11051114.
45. Radmacher M, et al. Measuring the viscoelastic properties of
human platelets with the atomic force microscope. Biophys
J 1996;70(1):556567.
46. Wu HW, Kuhn T, Moy VT. Mechanical properties of
L929 cells measured by atomic force microscopy: Effects of
anticytoskeletal drugs and membrane crosslinking. Scanning
1998;20(5): 389397.
47. Fisher TE, et al. The study of protein mechanics with the
atomic
force
microscope.
Trends
Biochem
Sci
1999;24(10):379384.
48. Altmann SM, Lenne P-F. Forced unfolding of single proteins.
49. Hafner JH, et al. Structural and functional imaging with
carbon nanotube AFM probes. Prog Biophys Molec Biol
2001;77(1):73110.
50. Yang Y, Wang H, Erie DA. Quantitative characterization of
biomolecular assemblies and interactions using atomic force
microscopy. Methods 2003;29(2):175187.
51. Sheng S, Zhifeng S. Cryo-atomic force microscopy. Methods
Cell Biol 2002;68:243256.
52. Fotiadis D, et al. Imaging and manipulation of biological
structures with the AFM. Micron 2002;33(4):385397.
53. Winfree E, et al. Design and self-assembly of two-dimensional
DNA crystals. Nature 1998;394(6693):539544.
54. Hansma HG, Kasuya K, Oroudjev E. Atomic force microscopy
imaging and pulling of nucleic acids. Curr Opin Struct Biol
2004;14(3):380385.
55. Jaeger L, Westhof E, Leontis NB. TectoRNA: Modular assembly units for the construction of RNA nano-objects. Nucl Acids
Res 2001;29:455463.
56. Seeman NC. DNA in a material world. Nature 2003;421:427
431.
57. Yan H, et al. DNA-templated self-assembly of protein arrays
and highly conductive nanowires. Science 2003;301(5641):
18821884.
58. Seeman NC. From genes to machines: DNA nanomechanical
devices. Trends Biochem Sci 2005;30(3):119125.
59. Radmacher M. Measuring the elastic properties of living cells
by AFM. Methods Cell Biol 2002;68:6790.
60. Langer MG, Koitschev A. The biophysics of sensory cells of
the inner ear examined by AFM and patch clamp. Methods
Cell Biol 2002;68:142171.
61. Charras GT, Lehenkari P, Horton M. Biotechnological applications of AFM. Methods Cell Biol 2002; 68.
62. Charras GT, Horton MA. Single cell mechanotransduction
and its modulation analyzed by atomic force microscope
indentation. Biophys J 2002;82(6):29702981.
See also BIOMAGNETISM;
NANOPARTICLES.

VIRGINIA M. AYRES
Michigan State University
East Lansing, Michigan
INTRODUCTION
Four years after its invention in 1982 (1), the scanning
tunneling microscope (STM) was awarded the 1986 Nobel
Prize for physics, one of only four such prestigious awards
given for a truly significant contribution to scientific
instrumentation. Since then, the family of scanning probe
microscopy (SPM) techniques, which includes scanning
tunneling microscopy, atomic force microscopy (24), magnetic force microscopy (5), near-field optical microscopy (6),
scanning thermal microscopy (7), and others, has revolutionized studies of semiconductors, polymers, and biological systems. The key capability of SPM is that, through a
controlled combination of feedback loops and detectors
with the raster motion of piezoelectric actuator, it enables
direct investigations of atomic-to-nanometer scale phenomena.
Scanning probe microscopy is based on a piezoelectricactuated relative motion of a tip versus sample surface,
while both are held in a near-field relationship with each
other. In standard SPM imaging, some type of tip-sample
interaction (e.g., tunneling current, Coulombic forces, magnetic field strength) is held constant in z through the use of
feedback loops, while the tip relative to the sample undergoes an xy raster motion, thereby creating a surface map
of the interaction. The scan rate of the xy raster motion
per line is on the order of seconds while the tip-sample
interaction is on the order of nanoseconds or less. The SPM
is inherently cable of producing surface maps with atomic
scale resolution, although convolution of tip and sample
artifacts must be considered.
Scanning tunneling microscopy is based on a tunneling
current from filled to empty electronic states. The selectivity induced by conservation of energy and momentum
requirements results in a self-selective interaction that
gives STM the highest resolution of all scanning probe
techniques. Even with artifacts, STM routinely produces
atomic scale (angstrom) resolution.
With such resolution possible, it would be highly desirable to apply STM to investigations of molecular biology
and medicine. Key issues in biology and medicine revolve
around regulatory signaling cascades that are triggered
through the interaction of specific macromolecules with
specific surface sites. These are well within the inherent
resolution range of STM.
The difficulty when considering the application of STM
to molecular biology is that biological samples are nonconductive. It may be more accurate to describe biological
samples as having both local and varying conductivities.
These two issues will addressed in this article, and
examples of conditions for the successful use of STM for
biomedical imaging will be discussed. We begin with an
overview of successful applications of STM in biology and
medicine.
517
SCANNING TUNNELING MICROSCOPY IN BIOLOGY AND

MEDICINE: DNA AND RNA
The STM imaging for direct analysis of base pair arrangements in DNA was historically the first biological application of the new technique. An amusing piece of scientific
history is that the first (and widely publicized) images (8
12) of (deoxyribonucleic acid) DNA were subsequently
shown to correspond to electronic sites on the underlying
graphite substrate! However, more careful investigations
have resulted in an authentic body of work in which the
base pairings and conformations of DNA and RNA are
directly investigated by STM. One goal of these investigations is to replace bulk sequencing techniques and crystal
diffraction techniques, which both require large amounts of
material, with the direct sequencing of single molecules of
DNA and RNA. Two examples of DNA and RNA investigation by STM are presented here. One is an investigation of
DNA and RNA structures, and the other is an investigation
of DNA biomedical function.
Recently reported research from the group at The Institute for Scientific and Industrial Research at Osaka University in Japan (13) has shown detailed STM images of
well-defined guanine-cytosine (G-C) and adenine-thymine
(A-T) base pairings in double- and single-stranded DNA.
Four simple samples involving only G-C and only A-T base
pairs in mixed (hetero) and single sided (homo) combinations were chosen for analysis (Fig. 1). These were deposited on a single-crystal copper (111)-orientation [Cu(111)]
substrate using a technique developed specially by this
group to produce flat, extended strands for imaging. An
STM image showing the individual A-T base pairs in the
hetero A-T sample is shown in Fig. 2. Images of the overall
structures indicated repeat distances consistent with interpretation as the double helix. Images from mixed samples
of hetero G-C and hetero A-T are shown in Fig. 3. The
larger structure is interpreted as hetero G-C and the
smaller as hetero A-T, which is consistent with X-ray
diffraction data that indicates the A-T combination is more
compact.
Only the double helix structure was observed for the
hetero G-C samples. However, the homo G-C structures,
(a)
(c)
Figure 2. STM image of portion of Hetero A-T double helix of

showing base pairs. (Reproduced from Ref. 9, used with
permission.)
hetero A-T structures, and homo A-T structures were

observed in two types, and the spot spacings and sizes of
the second type would be consistent with interpretation as
single-stranded DNA. The observed presence or lack of
single-stranded configurations among the samples is consistent with the fact that hetero G-C has a higher melting
(unraveling) temperature than the homo G-C and thus is
more difficult to unwind. Both hetero and homo A-T pairs
have lower melting temperatures than either of the G-C
pairs. Images of both hetero A-T and Homo A-T samples
often showed sizing and spacings consistent with interpretation as single-stranded DNA, in addition to observed
double helix specimens. Thus, the presence/lack of singlestranded versus double helix images is consistent with
known melting temperature data for the C-G and A-T base
pairings.
The same group has also reported successful STM
investigations of transfer-ribonuclic acid (t-RNA) (14). In
(b)
A
(d)
Figure 1. (a) Homo A-T, (b) Hetero A-T, (c) Homo G-C, and (d)
Hetero G-C. (Figure adapted from Ref. 9, used with permission.)
Figure 3. Hetero G-C and Hetero A-T mixed sample. The larger
specimens are identified as Hetero G-C, and the smaller specimens
are identified as Hetero A-T. Both are in a double helix
configuration. (Reproduced from Ref. 9, used with permission.)
518
(a)
TC arm
(b)
accepter stem
D arm
anticodon arm
6 nm
3
A
C
C
5 A
G C accepter stem
G C
G C
U A
C G
G C TC arm
U A
A
D arm
CGUCC U A
U
G
A
U GA
GCAGG UU C
CUCG
U
C
G
U
A
G
C
G
GUA
UGG
A
G U
extra loop
U A
U A
G C
A U anticodon arm
A
C
U
A
UUU
RNA, the base pairing is adenine-uracil (A-U) instead of

adenine-thymine (A-T). Also the backbone sugars are
ribose rather than deoxyribose, but are still linked by
phosphate groups. The RNA is very difficult to synthesize
as a single crystal and consequently there is a very limited
amount of X-ray diffraction data available for RNA. Little
is known about its variations, and therefore direct investigations of single molecule RNA would add much to our
knowledge.
Transfer RNA is a small RNA chain of 7493 nucleotides that transfers a specific amino acid to a growing
polypeptide chain at the ribosomal site of protein synthesis
during translation (15). It has sites for amino acid attachment, and an anticodon region for codon recognition that
binds to a specific sequence on the messenger RNA (mRNA) chain. It has a partial double-helix structure even
though it has only one chain, because the single RNA chain
folds back, and loops back, on itself, as shown in Fig. 4a.
X-ray diffraction studies (16) have indicated that the
t-RNA structure may often assume an L-shaped conformation with a long and a short arm. A model of the Escherichia
Coli lysine t-RNA macromolecule used by the group for its
STM studies is shown in Fig. 4a and b. It shows both the
L conformation and the underlying loop and base pair
chemistry.
Using STM, the group was able to directly image the L
conformation as shown in Fig. 4c. In addition to the first
direct statistical data on the lengths of the long and short
arms, obtained from analysis of several STM images, an
analysis of the influence of pH on conformation was also
carried out. Current investigations are focusing on biofunction research issues in addition to structural research
issues, using STM to directly image the coupling of the
important amino acid molecules at specific t-RNA sites.
The STM investigations of nanobiomedical rather than
structural issues are an important emerging research area.
One example is the recently reported research from the
University of Sydney group in which the local binding of
retinoic acid, a potent gene regulatory molecule, to plasmid
p-GEM-T easy (596 base pair Promega) DNA fragments on
a single-crystal graphite substrate, was directly imaged
and reported (17). Retinoic acid has been documented as
responsible for a number of profound effects in cell differentiation and proliferation, and is known to accomplish its
functions through selective site binding during the transcription process. The STM images of retinoic acid by itself
(c)
Figure 4. (a) Model of t-RNA L-shaped

conformation. (b) Model base pair arrangement
in L-shaped conformation. (c) STM image of Lshaped conformation at physiological pH.
(Reproduced from Ref. 10, used with permission.)
19.5 nm
on a single-crystal graphite substrate were investigated

first. These showed sizes consistent with the retinoic acid
molecular structure, and a bright head area with a darker
tail area. A molecular model of retinoic acid, also shown in
Fig. 5a, shows its aliphatic carbon ring head and polymeric
tail. For reasons further discussed below, the aliphatic ring
head may be expected to have a higher tunneling current
associated with it than the polymeric tail, and therefore the
observed bright and dark areas are consistent with the
expected structure.
At low concentrations, retinoic acid was observed to bind
selectively at minor groove sites along the DNA, with some
clustering of retinoic acid molecules observed, as shown in
Fig. 5b. High resolution STM imaging provided direct
evidence for alignment of the retinoic acid molecules
head-to-tail structure edge-on with the minor groove and
also in steric alignment with each other. From STM height
studies, it could also be inferred that the aliphatic ring
head was attached to a ring partner along the minor groove
surface, but that the tail was not attached. This may
suggest a loosely bound onoff functional mechanism. At
high concentrations, retinoic acid was observed to bind
along the whole length of the DNA double helix, but again
selecting the minor grooves. These first direct studies of
selective site binding of retinoic acid with the minor groove
of DNA should serve as a template for further direct
investigations of other known minor groove binders,
thereby opening up the direct investigation of an entire
Figure 5. (a) STM image of retinoic acid on a graphite substrate

compared with its molecular model showing the aliphatic ring
head and polymeric tail. (b) STM image of retinoic acid binding to
t-RNA with molecular model overlay. (Reproduced from Ref. 13,
used with permission.)
class of regulatory moleculeDNA interactions. The interactions of related structures that are candidate therapeutic
drug molecules could be receive similar direct investigation.
Note that both of the above groups have also made
important contributions to sample preparation techniques
for successful STM analysis of DNA and RNA. These
sample preparation techniques will be discussed below
in the context of the basic physics of the STM interaction,
and the basic chemistry and conductivity of DNA and RNA
samples.
BASIC PHYSICS OF THE STM INTERACTION
The STM is based on tipsample interaction via a tunneling
current between filled electronic states of the sample (or tip)
into the empty electronic states of the tip (or sample), in
response to an applied bias, as shown in Fig. 6. The bias may
be positive or negative, and different and valuable information may often be obtained by investigation of the how the
sample behaves in accepting, as well as in giving up, electrons. In STM imaging, it is important to recognize that the
feature map or apparent topography of the acquired image is
really a map of the local density of electronic states. Bright
does not correspond to a raised topography; it corresponds to
a region with a high density of electronic states. Therefore,
in STM imaging of biological samples, an important consideration is that a differential conductivity will be observed
from regions, such as rings (usually high) versus regions,
such as alkane backbones (usually low). As in all SPM
techniques, a z-direction feedback loop maintains some
aspect of the tip samples interaction constant (Fig. 6).
The readily available choices on commercial machines are
to hold either the tunneling distance d constant (constant
height mode) or the magnitude of the tunneling current
content (constant current mode).
The current in question is a tunneling current, which is
a quantum mechanical phenomenon. It is well documented
that all electrons within the atomic planes of any material
are in fact in such tight quarters that they display the
characteristics of a wave in a waveguide, in addition to
Feedback loop
Piezoelectric scanner
Controller
electronics
X, Y
Tunneling Current
Electron
energy is less
than barrier
energy
I ~ V expc d
0
Figure 7. A particle penetrating into and through a wall.
their particle-likeness. An electron at the surface of a

material faces a wall (barrier) created by the dissimilar
material (e.g., air, vacuum, or a liquid). While a particle
would run into a wall and bounce back, a wave can penetrate into and indeed through a wall (as light goes through
glass). This is illustrated in Fig. 7. Additionally, all materials have precise energy levels within them, and therefore,
electrons will move by going from one energy level at
location 0 to another at location d, meeting conservation
of energy requirements.
In STM, a tip with empty electronic states is brought
physically close to a sample surface. The electrons are
given a direction through the application of the bias (positive in this example). Because they are wavelike, when
they reach the sample surface, they can tunnel through the
barrier created by the 0-to-d gap and reach the empty
states of the tip, where they are recorded as a current
proceeding from sample to tip. A tunneling current has the
known mathematical form: I V expcd, where I is the
tunneling current, V is the bias voltage between the sample
and the tip, c is a constant and d is the tip-sample separation distance. The tunneling current depends sensitively
on the size of the 0-to-d gap distance. To observe a tunneling current, the gap must be on the order of tens of
nanometers. This is the case in any commercial STM
system. It is remarkable, that with the addition of a simple
feedback loop, a tip can be easily maintained within nanometers of a sample surface without touching it. Typical
STM tunneling currents are on the order of 1091012 A.
With special preamplifiers, currents on the order of 1014 A
can be detected.
Because STM is a current-based technique, some situations that can interfere with its current will be briefly
discussed. Very common in STM imaging of biological
samples is for the tip to acquire a layer of biological
material, possibly by going too close to the sample surface
while passing over an insulating region where the feedback
loop has little to work on. This usually just introduces
image artifacts, discussed below, but it can sometimes
insulate the tip from the sample, thus terminating the
tipsample interaction. The problem can be minimized
through careful consideration of the expected chemistry
and local conductivity of the biological specimen to be
investigated.
Tip
I ~ V expc d
519
Figure 6. Important features of an STM system.
CHEMISTRY, CONFORMATION, AND CONDUCTIVITY

OF BIOLOGICAL SAMPLES
Consideration of the basic chemistry involved in a biological sample can help to determine its appropriateness for
STM imaging. The building blocks for DNA and RNA are
520
(a)
O
CH2
(b)
H
H
H
O
NH2
Base
N
H
P
O
N
N
CH2
Base
NH
O
H
Base
N
H
GUANINE
O
NH2
URACIL
NH2
P
O
THYMINE
N
H
P
CH2
NH
NH
N
H
ADENINE
O
CH3
shown Fig. 8. The sugar-phosphate backbone contains

negatively charged phosphate groups for both DNA and
RNA. The bases adenine, thymine, uracil, guanine, and
cytosine are all nitrogenous ring systems. Thymine, cytosine, and uracil are six-member ring pyrimidine systems,
and adenine and guanine are purines, the fusion of a sixmember pyrimidine ring to a five-member imidazole ring.
Successful STM imaging of monolayers of the individual
bases has been reported (18,19). Examples of the high
resolution STM imaging that is possible for monolayers
of the individual bases are shown in Figs. 9 and 10.
The nitrogenous ring systems, like the classic benzene
ring system, which has also been imaged (20), contain porbital electrons above and below the ring structure plane,
which create a conductive electron cloud. Hence, the successful STM imaging of the DNA and RNA systems by the
Osaka University and University of Sydney groups might
be expected from the charged phosphate groups in the
backbones and the ring systems in the base pairs.
However, there are also very difficult issues to resolve in
making the local conductivity of, especially, the signature
DNA and RNA base pairs available to the STM tip. These
are enclosed within the sugar-phosphate backbones, and
only partially exposed by the twisting of the helix, as shown
in Fig. 11a and b (21,22). Also, the choice of substrate will
powerfully influence the molecular structure deposited on
it, especially if it is small. An example of this is shown in
Fig. 12, taken from Ref. 16. The behaviors of pyridine (a
Figure 9. STM images of (a) guanine, (b) cytosine, and (c) adenine
monolayers on a single crystal (111)-orientation gold substrate.
(Reproduced from Ref. 14, used with permission.)
N
N
H
CYTOSINE
O
Figure 8. (a) The deoxyribose (ribose) sugar/
phosphate backbone for DNA (RNA) is
negatively charged due to phosphate groups.
(b) DNA and RNA bases are nitrogenous ring
systems.
single-nitrogen close relation to pyrimidine) and benzene

on a single crystal (001) orientation copper, Cu(001), substrate were investigated. The pyrimidine monolayers (thymine, cytosine, and uracil) in Figs. 9 and 10 had rings
oriented parallel to the substrate surface, but individual
pyridine molecules on Cu(001) had rings perpendicular to
the surface, due to the strong nitrogen-copper atom interaction, as shown in Fig. 12a. Also, if a single hydrogen atom
was dissociated from the pyridine molecule, as can happen
during routine scanning, the molecule would shift its position on the copper substrate (Fig. 12b). The STM imaging of
an individual benzene molecule indicated a ring system
parallel to the copper substrate (Fig. 12c), but hydrogen
dissociation would cause the benzene molecule to become
perpendicular to the substrate surface (Fig. 12d). Therefore both the substrate choice and interactions with the
imaging tip can influence the conformation of the biomolecule and whether its locally conductive portions are
positioned to produce a tunneling current.
Now consider the situation of a molecule with a difference in local conductivity, like retinoic acid. The aliphatic
ring head would similarly be expected to have a high local
conductivity, and separate investigations of just retinoic
acid by the University of Sydney group confirmed that this
is the case (Fig. 5a). The polymeric tail is basically an
alkane system without any p-type orbitals. Its conductivity
is therefore expected to be less than the ring system and
this is experimentally observed. However, results such as
those shown in Fig. 13 from a group at California Institute
of Technology, demonstrate that high resolution STM
imaging even of low conductivity alkane systems is possible
(2326). Therefore, one aspect of STM biomolecular imaging is that there may be large differences in the conductivities of two closely adjacent regions. It then becomes an
issue of whether the STM feedback loop will be able to
sufficiently respond to the differences to maintain the tipsample tunneling current interaction throughout the
investigation. Prior consideration of the imaging parameters necessary for successful STM imaging of the least
conductive part of the bio molecule can help.
521
Figure 10. STM images of (a) guanine, (b)

adenine, (c) uracil, and (d) thymine monolayers
on (e) a single crystal (0001)-orientation
molybdenum dissulfide substrate. (Adapted
from Ref. 15, used with permission.)
Biomolecules, with only nanometer dimensions, always

should be deposited on atomically flat single-crystal substrates. Substrates can also be selected to supply electrons
to the biomolecule, for positive bias scanning, or to manipulate the biomolecule into a desired position. Another
important sample preparation issue is that biomolecules
often have multiple available conformations, including
globular conformations that self-protect the molecule
under nonphysiological conditions. While STM imaging
may be performed in vacuum, air, and even in a liquidfilled compartment (liquid cell), the best resolution may be
achieved in vacuum, which is a nonphysiological condition.
The less physiological the imaging conditions, the more it
will be necessary to use special molecular stretching techniques to investigate an open conformation. A special
pressure jet injection technique was developed by the
Osaka University group to deposit stretched DNA and
RNA on single-crystal copper for vacuum STM imaging,
without giving them the chance to close into globular
conformations (13,14).
Figure 11. The three-dimensional conformation of DNA. (a) The
base pairs are positioned between the sugar-phosphate backbones.
(b) The overall structure is a double helix. (Reproduced from Refs.
17,18, used with permission.)
(a)
Pyridine on Cu(001)
(b)
(c)
(d)
Benzene on Cu(001)
2H
Figure 12. Influence of the sample-substrate interaction on

sample orientation. (a) An individual pyridine molecule on a
copper (001)-orientation, (Cu(001)) substrate is perpendicular to
the surface due to the strong nitrogencopper atom interaction. (b)
An individual pyridine molecule from which a hydrogen atom has
dissociated is also perpendicular to a Cu(001) surface but has a
shifted location. (c) An individual benzene molecule on a Cu(001)
substrate is parallel to the surface but (d) may become
perpendicular if hydrogen dissociation occurs. (Adapted from
Ref. 16, used with permission.)
Figure 13. High resolution STM images of an alkane

(pentatracontane) monolayer on graphite. (Reproduced from
Ref. 19, used with permission.)
522
IMAGING ARTIFACTS AND DATA RESTORATION

USING DECONVOLUTION
Examination of Fig. 5a shows the ring head of retinoic acid
as a large blurred bright spot. Greater resolution of detail
would clearly be desirable. As in all SPM imaging systems,
tip artifacts versus the surface features will limit the
resolution of the experiments performed. This is often cited
as an ultimate barrier in STM studies of macromolecular
structures and in scanning probe microscopy in general
(27). It is therefore necessary to develop techniques for
deconvolution of STM tip artifacts for enhancing the resolution of measured STM image data.
A commonly used approach for data restoration or
eliminating the smearing effect of tip sample interaction
is to assume that the observed signal is a convolution of the
true image and the probe response function (PRF). The
following equation gives a general degradation model due
to the convolution of tip artifacts with true data resulting in
the measurement g(x,y). Neglecting the presence of the
additive noise, the data can be modeled as
X
gx; y f x; y hx; y
f n; mhx n; y m
n;m
where g(x, y), f(x, y), and h(x,y) are the observed or raw
signal, true image, and PRF, respectively. One can then
use deconvolution methods to extract the true image from
the knowledge of measured data and probe PRF.
Theoretically, the probe response function is derived
from the underlying physics of the tip sample interaction
process. Hence, there is a need for a theoretical model for
the tip sample interaction. Recent advances in formulation
and modeling of tip sample interactions allow development
of accurate compensation algorithms for deconvolving the
effect of tip-induced artifacts.
Figure 14 shows an example of applying a deconvolution
algorithm on synthetic degraded images. The degraded
image in Fig. 14c is generated from a synthetic image in
Fig. 14a blurred by a Gaussian PRF in Fig. 14b. Figure 14d
shows the enhanced result obtained using deconvolution.
Although the theoretical treatment of STM and related
SPM techniques provide major challenges because the
atomic structures of the tip and sample have to be modeled
appropriately, its potential is clear and this is a strongly
developing research area at the present time.
CONCLUSIONS
The STM imaging has the highest resolution of all SPM
imaging techniques. As such, it would be highly desirable
to apply STM to investigations of molecular biology and
medicine. An often described difficulty when considering
the application of STM to molecular biology is that biological samples are nonconductive. It would be more accurate
to describe biological samples as having both local and
varying conductivities. Design of STM experiments in
which ring systems are exploited, and/or imaging parameters are set for the least conductive portion of the
biomolecules may help produce successful imaging results.
New research in applications of powerful deconvolution
Figure 14. Clockwise from upper left: (a) synthetic true image
(b) Gaussian PRF, (c) degraded measurement, and (d) restored
image.
techniques to STM imaging will also open up the field of

direct STM investigations of the structure and function of
important biomolecules.
BIBLIOGRAPHY
1. Binning G, Rohrer H. Helv Phys Acta 1982;55:726735.
2. Hansma HG, Oroudjev E, Baudrey S, Jaeger L. TectoRNA and
kissing loops: Atomic force microscopy of RNA structures. J
Microsc 2003;212:273279; Sitko JC, Mateescu EM, Hansma
HG. Sequence-dependent DNA condensation and the electrostatic zipper. Biophys J 2003;84:419431; Hansma HG,
Vesenka J, Siegerist C, Kelderman G, Morrett H, Sinsherimer
RL, Bustanmante C, Elings V, Hansma PK. Reproducible
imaging and dissection of plasmid DNA under liquid with
the atomic force microscope. Science 1992;256:1180.
3. Hartmann U. Magnetic force microscopy. Annu Rev Mater
Sci 1999;29:5387.
4. Paesler MA, Moyer PJ. Near-Field Optics: Theory, Instrumentation, and Applications. New York: Wiley-Interscience; 1996.
5. Majumdar A. Scanning thermal microscopy. Annu Rev Mater
Sci 1999;29:505585.
6. Beebe TP, Jr., et al. Science 1989;243:370371.
7. Lee G, Arscott PG, Bloomfiled VA, Evans DF. Science
1989;244:475478.
8. Driscoll RJ, Youngquist MG, Baldescwieler JD. Atomic scale
imaging of DNA using scanning tunnelling microscopy.
Nature (London) 1990;346:294296.
9. Tanaka H, Kawai T. Visualization of detailed structures
within DNA. Surface Sci Lett 2003;539:L531L536.
10. Nishimura M, Tanaka H, Kawai T. High resolution scanning
tunneling microscopy imaging of Escherichia coli lysine transfer ribonucleic acid. J Vac Sci Technol B 2003;21:12651267.
11. Wikipedia, The free encyclopedia. Available at http://en.wikipedia.org/wiki/RNA.
MICROSURGERY
12. Giege R, Puglisi JD, Floentz C. In: Cohn WE, Moldave K, eds.,
Progressin Nucleic Acid Research and Molecular Biology,
Vol. 45. Amsterdam: Elsevier; 1993.
13. Hadi Zareie M, Lukins PB. Atomic-resolution STM structure
of DNA and localization of the retinoic acid binding site.
Biochem Biophy Res Commun 2003;303:153159.
14. Otero R, et al. Proceedings of the 5th Trends In Nanotechnology (TNT04) CMP Cientifica; 2004. Available at http://
www.phantomsnet.net/files/abstracts/TNT2004/
AbstractKeynoteBesenbacherF.pdf.
15. Available at http://biochem.otago.ac.nz/staff/sowerby/periodicmonolayers.htm. Multiple references listed.
16. Lauhon LJ, Ho W. Single molecule chemistry and vibrational
spectroscopy: Pyridine and benzene on Cu(001). J Phys Chem
A 2000;104:24632467.
17. Image credit in Fig. 11(a): U.S. Department of Energy
Human Genome Program, Available at http://www.ornl.gov/hgmis. This image originally appeared in the 1992
U.S. DOE Primer on Molecular Genetics.
18. Image credit in Fig. 11(b): Mathematical Association of America, Available at http://www.maa.org/devlin/devlin0403.
html.
19. Claypool CL, et al. Source of image contrast in STM images of
functionalized alkanes on graphite: A systematic functional
group approach. J Phys Chem B 1997;101:59785995.
20. Faglioni F, Claypool CL, Lewis NS, Goddard WA III. Theoretical description of the STM images of alkanes and substituted alkanes adsorbed on graphite. J Phys Chem B
1997;101:59966020.
21. Claypool CL, et al. Effects of molecular geometry on the STM
image contrast of methyl- and bromo-substituted alkanes
and alkanols on graphite. J Phys Chem B 1999;103:9690
9699.
22. Claypool CL, Faglioni F, Goddard WA III, Lewis NS. Tunneling mechanism implications from an STM study of
H3C(CH2)15HC C CH(CH2)15CH3 on graphite and
C14H29OH on MoS2. J Phys Chem B 1999;103:70777080.
23. Villarubia JS. Algorithms for scanned probe microscope
image simulation, surface reconstruction and tip
estimation. J Res Nat Inst Standards Technol 1997;102:
425454.
See also BIOSURFACE
ENGINEERING; MICROSCOPY, ELECTRON.
MICROSENSORS FOR BIOMEDICAL

APPLICATIONS. See CAPACITIVE MICROSENSORS FOR
BIOMEDICAL APPLICATIONS.
MICROSURGERY
KEITH J. REBELLO
The Johns Hopkins University
Applied Physics Lab
Laurel, Maryland
INTRODUCTION
Microsurgery is a specialized surgical technique whereby a
microscope is used to operate on tiny structures within the
body. Fine precision microinstruments are used to manipulate tissue with or without robotic or computer control.
523
This technique has allowed for significant advances in

surgery especially for operations involving the inner ear,
eye, brain, nerves, and small blood vessels.
HISTORY OF THE SURGICAL MICROSCOPE

The compound microscope is generally agreed upon to have
been invented in the 1590s by the two Dutch opticians
Hans and Zacharias Janssen. Their device consisted of a
sliding tube with two aligned lenses. In 1624 Galileo
Galilei, the famous astronomer and mathematician,
demonstrated an inverted telescope to his colleagues of
the Lincean Academy. One of them, Giovanni Faber,
named it a microscope from the Greek words micro, meaning small, and scope, meaning to aim or shoot. Microscope
lenses were improved in the seventeenth century by
Antonie van Leeuwenhoek, a Dutch linen draper who
originally was interested in counting the number of
threads per square inch in his reams of cloth. How he
constructed his spherical lenses still remains a mystery
to this day. In the eighteen century, fine and course adjustments as well as tube inclination were added by Robert
Hooke, who first discovered the cell. The microscope was
further improved by Joseph Jackson Lister a British wine
merchant, school visitor, and histologist, the father of Lord
Joseph Lister whom is credited as stating the era of modern
surgery. His innovations included the developed achromatic objective lens corrected for chromatic and spherical
aberrations and stands designed to reduce vibrations. His
jointly published work with Dr. Joseph Hodgkins in 1827,
redefined the understanding at the time of arteries, muscles, nerves, and the brain.
In 1848 Carl Zeiss, a German machinist opened a microscope workshop. Ernst Abbe , a physicist working with
Zeiss, derived new mathematical formulas and theories
that allowed the optical properties to be mathematically
predicted for the first time. Prior lenses had always been
made by craftsmen who learned their trade by trial and
error. Abbe s advancements enabled Zeiss to become the
first mass producer of high quality microscopes.
USE IN SURGERY
Edwin Theodor Saemisch, a German ophthamologist, used
loupes in surgery in 1876, but although the microscope was
being used in the laboratory medical research environment
it was not used the operating room. Zeiss manufactured a
binocular microscope specifically designed for dissecting
which was used for ophthalmological examinations of the
cornea and anterior chamber of the eye. It was not until
1921 that Carl Olof Nylen, a Swedish, otologist and tennis
olympian, used his homebuilt monocular microscope for
the first time in ear surgery on a case of chronic otis media,
a type of ear infection. His monocular microscope was
quickly replaced in 1922 by a binocular microscope developed by adding a light source to a Zeiss dissecting microscope by his chief surgeon, Gunnar Holmgren. He used it to
treat diseases otosclerosis, the abnormal growth of temporal bone in the middle ear.
524
MICROSURGERY
Despite these early successes the surgical microscope

was seldom used due to its limited field of vision, very short
focal distance, poor light quality, and instability. It was not
until the 1950s that the surgical microscope started to
become more widely adopted. In 1953, Zeis released the
Zeiss OpMi 1(Zeiss Operating Microscope Number One),
which was specially designed for otological procedures. Its
superior coaxial lighting, stability, and ease of operation
enabled the advent of tympanoplasty operations to repair
ruptured ear drums as well as widespread use in temporal
bone surgery.
The success the microscope was having in otology soon
spread to other disciplines as well. In the early 1950s, Jose
Ignacio Barraquer adapted a slip lamp to the Zeiss otological surgical microscope for ocular microsurgery. By the
1960s, J. I. Barraquer, Joaqun Barraquer, and Hans Littman of Zeiss, had further modified the surgical microscope
and refined microsurgical techniques to make ocular maneuvers in glaucoma microsurgery easier to perform. During
this same time frame, Richard Troutman also had Zeiss
make a special microscope for his ophthalmic procedures.
He made many advances and is credited as adding electric
and hydraulic control to surgical microscopes, but possibly
his greatest innovation was the first variable magnification, or zoom, surgical microscope.
Around this time neurosurgeons also began using the
surgical microscope in the operating room. In 1957, Theodore Kurze removed a neuriloma tumor from the seventh
cranial nerve, and then later anastomized it to the hypoglossal nerve. He also developed techniques to use the
surgical microscope for aneurysm surgeries. Recognizing
that sterilization was a major problem, he developed the
use of ethylene oxide gas to sterilize his surgical microscopes for use in the operating room. Dr. R. M. Peardon
Donaghy established the first microsurgical training lab at
the University of Vermont, where many surgeons were
trained. He collaborated with the vascular surgeon Julius
Jacobson to remove occlusions from cerebral arteries.
Jacobson and his colleague Ernesto Suarez, were responsible for developing small vessel anastomoses techniques.
Their procedures required another surgeons assistance.
To meet this need Jacobson invented the diploscope that
allowed two surgeons to view the same operative field.
Later he and his colleagues worked with Hans Littman
of Zeiss to develop a commercial surgical microscope with a
beamsplitter enabling two surgeons to operate at the same
time. A modern day version of the Zeiss dual head microscope is shown in Fig. 1.
Inspired by the work of the neuroscientists plastic surgeon also started using the surgical microscope. Harold
Buncke was one of the first plastic surgeons to use the
microscope for digit/limb replantation and free-flap autoplantation. Buncke also developed many of the tools microsurgeons use by borrowing technology from the jewelry,
watchmaking, and microassembly industries (Fig. 2.)
The 1960s saw the techniques applied to neurosurgery.
Breakthroughs in microneurosurgery included the repair
of peripheral nerve injuries, intracranial aneurysm surgeries, embolectomies of middle cerebral arteries, middle
cerebral artery bypasses. One on the visionaries of this
time was M. G. Yasargil a Turkish neurosurgeon from
Figure 1. Zeiss OpMi Vario S8 surgical microscope. (Courtesy of

Carl Zeiss.)
Switzerland. Trained in Donaghys training lab he further

refined and improved the surgical techniques.
The next advancements came with the development of
minimally invasive surgical techniques. Traditional surgical techniques used relatively large incisions to allow the
surgeon full access to the surgical area. This type of operation, called open surgery, enables the surgeons hands and
instruments to come into direct contact with organs and
tissue, allowing them to be manipulated freely. These
operations are classified as first generation surgical techniques and most surgeons are trained in this manner.
While the large incision gives the surgeon a wide range
of motion to do very fine controlled procedures, it also
causes substantial trauma to the patient. In fact, the
majority of trauma is caused by the incisions the surgeon
uses to get access to the surgical site instead of the actual
Figure 2. Modern day microsurgical tools. (Courtesy of WPI Inc.)
MICROSURGERY
surgical procedure itself. For example, in a conventional

open-heart cardiac operation, the rib cage must be cracked
and split exposing the heart muscle. This trauma not only
increases pain to the patient, but adds to recovery times
increasing hospital stays, in turn increasing costs.
In 1985, Muhe performed the first laparoscopic cholecystectomy, or gallbladder removal surgery with a fiberoptic scope. The technique he performed is commonly
called minimally invasive surgery but also goes by other
names, such as micro, keyhole, microscopic, telescopic, less
invasive, and minimal access surgery. This microsurgical
technique is based on learnings from gynecological pelviscopies and arthroscopic orthopedic operations along with
the previous advances made in otology, opthamology, neurosurgery, and reconstructive microsurgeries. It has subsequently been applied to many other surgical areas, such
as general surgery, urology, thoracic surgery, plastic surgery, and cardiac surgery. These procedures are classified
as second generation surgeries as trauma to the patient is
drastically reduced by the reducing or eliminating incisions. The shorter hospital stays and faster recovery times
for the patient reduce the cost of a minimally invasive
procedure 35% compared to its open surgery counterpart.
In a minimally invasive cardiac operation a few small
holes, access points, or ports are punctured into the patient
and trocars are inserted. A trocar consists of a guiding
cannula or tube with a valveseal system to allow the body
to be inflated with carbon dioxide. This is done so that the
surgeon has enough room to manipulate his instruments at
the surgical site. An endoscope is inserted into one of the
trocar ports to allow the surgeon a view the surgical site.
Various other surgical instruments, such as clippers, scissors, graspers, shears, cauterizers, dissectors, and irrigators were miniaturized and mounted on long poles so that
they can be inserted and removed from the other trocar
ports to allow the surgeon to perform the necessary tasks at
hand.
While minimally invasive surgery has many advantages
to the patient, such as reduced postoperative pain, shorter
hospital stays, quicker recoveries, less scarring, and better
cosmetic results, there are a number of new problems for
the surgeon. The surgeons view is now restricted and does
not allow him to see the entire surgical area with his eyes.
While the operation is being performed he must look at a
video image on a monitor rather than at his hands. This is
not very intuitive and disrupts the natural handeye coordination we all have been accustomed to since childhood.
The video image on the monitor is also only two dimensional (2D) and results in a loss of our binocular vision
eliminating the surgeons depth perception. While performing the procedure the surgeon does not have direct
control of his own field of view. A surgical assistant holds
and maneuvers the endoscopic camera. The surgeon has to
develop his own language to command the assistant
to position the scope appropriately, which often leads to
orientation errors and unstable camera handling, especially during prolonged procedures. Since the images from
the camera are magnified, small motions, such as the
tremor in a surgical assistants hand or even their heartbeat can cause the surgical team to experience motion
induced nausea. To combat the endoscopic problems, some
525
surgeons choose to manipulate the endoscope themselves.

This restricts them to using only one hand for delicate
surgical procedures and makes procedures even more
complicated.
The surgeon also loses the freedom of movement he has
in open surgery. The trocar ports are fixed to the patients
body walls by pressure and friction forces. This constrains
the instruments motion in two directions and limits the
motion of the tip of the instrument to four degrees of
freedom (inout, leftright, updown, and rotation). The
trocars also act as pivot points and cause the surgical
instruments to move in the opposite direction to the surgeons hands. When the surgeon is moving left, the image
on the monitor is moving to the right. The amount of this
opposite movement also depends on the depth of the introduction of the instrument. Again because of the pivot point
the deeper an instrument is inserted into the body the more
the surgeons movement is amplified. Even a small movement made by the surgeon on the outside of the patient can
translate to a very large movement on the inside of the
patient. The seals and valves in the trocars also impede
movements which hinders the smoothness of motions into
and out of the patient and greatly reduces the already
limited tactile feedback the surgeon experiences. These
movement behaviors and lack of tactile feedback are counter to what the surgeon is used to in open surgery and
require long training to develop the technical skills to
perform these operations.
Performing a minimally invasive procedure has been
likened to writing your name holding the end of an 18 in.
(45.72 cm) pencil (1). The surgeon loses three-dimensional
(3D) vision, dexterity, and the sense of touch. The instruments are awkward, counterintuitive, and restricted in
movement. The lack of tactile feedback prevents the surgeon from knowing how hard he or she is pulling, cutting,
twisting, suturing, and so on. These factors cause a number
of adjustments to be made by the surgeon, which requires
significant retraining on how to do the procedures in a
minimally invasive manner. These difficulties encountered
by the surgeon cause degradation in surgical performance
compared to open surgery which limits surgeons to performing only simpler surgical procedures.
In an attempt to address some of these shortcomings
and allow the surgeon more control during operations a
third generation of surgical procedures, robotic surgery
was developed. Although these types of procedures are
commonly referred to as robotic surgery, the operations
themselves are not completely automated and are still
carried out by a surgeon. For this reason, robotic surgery
is also referred to as computer aided or computer assisted
surgery.
The robotic technology was originally developed for
telerobotic applications in the late 1980s for the Defense
Advanced Research Project Administration (DARPA) by
researchers at SRI International. The surgeon of the future
would allow surgeons from remote command centers to
operate on injured soldiers in the battlefield. In 1995, this
technology was spun off into a new company named Intuitive Surgical to commercialize the technology for use in the
hospital environment. Near the same time Dr. Yulan Wang
was developing robotic technology for NASA to allow
526
MICROSURGERY
Figure 3. Intuitive Surgical da Vinci robotic surgery system.

(Copyright #2005 Intuitive Surgical, Inc.)
surgeons on earth to deal with medical emergencies on the

international space station. He formed Computer Motion
in 1989. Both of these companies merged in 2003, and
Intuitive Surgical is now the leader in Robotic Surgery.
In 2002, Dr. Fredric Mol, one of the original founders
of Intuitive Surgical, founded Hansen Medical which
brings computerized robotic control of catheters to electrophysiology and interventional cardiac procedures.
Current robotic surgery systems have a number of benefits over conventional minimally invasive surgery. Figure 3
shows an Intuitive Surgical da Vinci robotic system. In this
arrangement the surgeon sits comfortably at a computer
console instead of having to stand throughout the entire
procedure, which can last up to 5 h long. A three-armed
robot takes his place over the patient. One arm holds an
endoscope while the other two hold a variety of surgical
instruments. The surgical team can also look at a video
monitor to see what the surgeon is seeing. The surgeon looks
into a stereo display in much the same way as looking
though a surgical microscope and manipulates joystick
actuators located below the display. This simulates the
natural handeye alignment he is used to in open surgery,
(Fig. 4). Since computers are used to control the robot and
are already in the operating room, they can be used to give
the surgeon superhuman like abilities. Accuracy is
improved by employing tremor cancellation algorithms to
filter the surgeons hand movements. This type of system
can eliminate or reduce the inherent jitter in a surgeons
hands for operations where very fine precise control is
needed. Motion scaling also improves accuracy by translating large, natural movements into extremely precise, micromovements. A wide variety of surgical instruments or end
effectors are available including graspers, cutters, cauterizers, staplers, and so on. Both companies provide end
effectors that have special wrist like joints at their tips
enabling full seven degree of freedom movements inside
the patient, (Fig. 5), but still lack tactile feedback.
These robotic advances allow surgeons to perform more
complex procedures, such as reconstructive cardiac operations like coronary bypass and mitral valve repair that cannot
be performed using other minimally invasive techniques.
Figure 4. Intuitive Surgical stereo display and joysticks.

(Copyright #2005 Intuitive Surgical, Inc.)
Figure 5. Multidegrees-of-freedom end effector. (Copyright

#2005 Intuitive Surgical, Inc.)
MICROSURGERY
527
MEMS
Around the same time that minimally invasive surgery was
being developed, there was a turning point in microelectromechanical systems (MEMS). This a technology was
developed from the integrated circuit industry to create
miniature sensors and actuators. Originally, these semiconductor processes and materials were used to build
electrical and mechanical systems, but have now expanded
to include biological, optical, fluidic, magnetic, and other
systems as well. The term MEMS originated in the United
States and typically contain a moving or deformable object.
In Europe, this technology goes by the name microsystems
technology or microstructures technology (MST) and also
encompasses the method of making these devices, which is
referred to as micromachining. In Japan and Asia MEMS
are called micromachines when mechanisms and motion
are involved. The MEMS devices first were used in medical
applications in the early 1970s with the advent of the
silicon micromachined disposable blood pressure sensor
(2), but it was not until the mid-1980s when more complicated mechanical structures, such as gears and motors,
were able to be fabricated.
FABRICATION TECHNOLOGIES
The fabrication of MEMS devices is based on the merger of
semiconductor microfabrication processes and micromachining techniques to create the desired microstructural
components. There are four major processes that are used
to fabricate MEMS devices: bulk micromachining, surface
micromachining, LIGA, and precision machining. Combinations of these technologies are what allow MEMS to be
highly miniaturized and integratable with microelectronics. These processes are very sensitive to impurities and
environmental conditions such as temperature, humidity,
and air quality. Typically, these fabrication steps are performed inside a cleanroom (Fig. 6). Bulk micromachining,
surface micromachining, and LIGA have the added advantage of being able to be batch fabricated. This allows many
devices to be made in parallel at the same time greatly
reducing device cost.
Bulk micromachining utilizes wet- or dry-etch processes
to form 3D structures out of the substrate. These subtractive processes produce isotropic or anisotropic etch profiles
in material substrates, which are typically but not limited
to silicon wafers. Bulk micromachining can create large
MEMS structures on the micrometers (mm) to millimeters
(mm) scale (tens of mm-to-mm thick). Commercial applications of bulk micromachining have been available since the
1970s. These applications include pressure sensors, inertial sensors such as accelerometers and gyros, and microfluidic channels and needles for drug delivery.
In surface micromachining, MEMS are formed on the
surface of the substrate using alternating layers of structural and sacrificial materials. These film materials are
repeatedly deposited, patterned, and etched to form structures that can then be released by removing sacrificial
layers. The release process allows for the fabrication of
complex movable structures that are already assembled,
Figure 6. Cleanroom fabrication facility. (Courtesy of Intel Corp.)
such as motors, switches, resonators, cantilevers, and so

on. Surface micromachined structures are typically limited
to thicknesses of 26 mm and because they use much of the
same technology as is used in the integrated circuit industry are readily integrated with electronics. Because so
much technology is shared with the IC industry silicon
wafers are typical substrates with thousands of devices
being able to be fabricated at once (Fig. 7).
Lithographie, Galvanik, Abformung (LIGA) is a German
acronym that means lithography, electroforming, and
molding. The technology was originally developed in
the late-1970s to fabricate separation nozzles for uranium
enrichment. This technology uses X rays to fabricate
devices with very high aspect ratios. A synchrotron radiation source is used to define small critical dimensions in a
poly(methyl methyacrylate) (PMMA), mold that can then
be electroplated to form high aspect ratio metallic structures. Many parts can be batch fabricated in this manner,
but assembly is usually still a serial process.
Precision machining technology, such as micro-EDM
(microelectro discharge machining), laser micromachining,
and micro stereo lithography, is also used to form complex
structures out of metal, plastic, and ceramics that the
previous fabrication technologies may be incapable of.
Precision machining is typically a serial process, but is
528
MICROSURGERY
Figure 8. Strain gauges fabricated on surgical sharps. (Courtesy

of Verimetra, Inc.)
Figure 7. Silicon wafer in the fabrication process. (Courtesy of

Intel Corp.)
often better able to deal with the varied shapes and substrates of microsurgical instruments.
Micro EDM is a form of spark machining used to shape
conductive materials, such as silicon and metals. An
EDM erodes material by creating a controlled electric
discharge between an electrode and the substrate. It is
a noncontact process and there is no direct mechanical
cutting force applied to the substrate. Dielectric fluid is
used to remove the erosion particles, as well as to keep
the substrate material from oxidizing. Micro-EDMs can be
used to make holes, channels gears, shafts, molds, dies,
stents, as well as more complex 3D parts such as accelerometers, motors, and propellers (3).
Lasers can be used to both deposit and remove material.
Laser ablation vaporizes material through the thermal
noncontact interaction of a laser beam with the substrate.
It allows for the micromachining of silicon and metals, as
well as materials that are difficult to machine using other
techniques such as diamond, glass, soft polymers, and
ceramics. Laser direct writing and sintering is a maskless
process where a laser beam is used to directly transfer
metal materials onto a substrate. This can be used to form
metal traces on nonplaner surfaces, which reduces the
need for wires on surgical tools (4).
Micro stereo lithography processes generate 3D structures made out of ultraviolet (UV) cured polymers. It is an
additive process where complex 3D structures are made
from stacks of thin 2D polymer slices that have been
hardened from a liquid bath. Conventional systems were
limited in that they were a serial process where only one
part could be made at a time. MicroTEC has developed a
batch fabricated wafer level process called rapid material
product development (RMPD), which is capable of constructing structures out of 100 different materials including plastics, solgels, and ceramics (5).
only overcome many of the limitations of microsurgical

procedures, but allow for new more advanced operations
to be performed. MEMS are just now being incorporated
into microsurgical tools and coming on the market. Most
are still at the research level, but the industry is moving in
this direction as the need for smaller smarter tools
increases.
HAPTIC FEEDBACK
One of the key areas for improvement in microsurgery is
tactile feedback. The lack of tactile sensing limits the
effectiveness these procedures. Recent work in robotic
feedback for minimally invasive surgery has concentrated
on force feedback techniques using motors and position
encoders to provide tactile clues to the surgeon. In these
approaches, the sense element is far removed from the
sense area. Verimetra, Inc. has developed strain gauge
force sensor fabrication technology which uses the surgical
tools themselves as a substrate (6). Prior efforts have
focused on fabrication of sensors on silicon, polyimide, or
some other substrate followed by subsequent attachment
onto a surgical tool with epoxy, tape, or some other glue
layer. Attaching a sensor in this manner limits performance, introduces sources of error, limits the sensors size,
and further constrains where the sensor can be placed. By
eliminating glue and adhesion layers improved sensitivity
and reduces errors due to creep. Figure 8 shows strain
gauges fabricated on surgical sharps. Figure 9 is a cut away
SEM image of a strain gauge and temperature sensor
embedded inside of a robotic microforcep. While this microfabrication technology is an improvement in sensor technology, wires are still used to connect the sensor to the
outside world. Reliability and the added complexity of
adding wires to surgical end effectors with high degrees
APPLICATIONS
The inclusion of MEMS technology in microsurgery, will
allow for smaller more miniaturized surgical tools that not
Figure 9. Strain gauges and temperature sensor embedded in

robotic microgripper. (Courtesy of Verimetra, Inc.)
MICROSURGERY
of freedom limit the effectiveness of the technology. Shortrange wireless technology compatible with the operating
room environment need to be developed to overcome these
limitations.
Recently tactile feedback has been shown to be able to be
added to noncontact lasers. Based on optical distance
measurements, the systems synthesize haptic feedback
through a robotic arm held by the surgeon when the focal
point of the laser is coincident with a real surface. This
gives the operator the impression of touching something
solid. By increasing the power of the laser such a system
could also be used for cutting or ablation.
TISSUE SENSING
Taking haptic feedback one step further is the ability to
distinguish between different types of tissue in the body.
Tissue sensing is of vital importance to a surgeon. Before
making an incision into tissue, the surgeon must identify
what type of tissue is being incised, such as fatty, muscular,
vascular, or nerve tissue. This becomes more complicated
because the composition and thickness of different human
tissues varies from patient to patient. Failure to properly
classify tissue can have severe consequences. For example,
if a surgeon fails to properly classify a nerve and cuts it,
then the patient can suffer effects ranging from a loss of
feeling to loss of motor control. If a neurosurgeon cuts into a
blood vessel while extracting a tumor severe brain damage
may occur. The identification and classification of different
types of tissue during surgery, and more importantly during the actual cutting operation, will lead to the creation of
smart surgical tools. If a surgical tool senses that it is too
close to or about to cut the wrong type of tissue it can simply
turn itself off.
Verimetra, Inc. has developed a device called the data
knife, Fig. 10. It is a scalpel, which is outfitted with
different strain sensors along the edges of the blade to
sense the amount of force being applied. The resistance of
the tissue is one of the signals used for classifying tissue.
Pressure sensors are used to measure the characteristics of
material surrounding the blade. The pressure of the surrounding fluid can be used to help classify the type or
location of tissue. Electrodes are used to measure the
impedance of different types of tissue, as well as being
used to identify nerves by picking up their electrical signals. The tool provides the real-time feedback surgeons
Figure 10. Data Knife smart scalpel. (Courtesy of Verimetra,

Inc.)
529
have been asking for during surgery, and can also be used
to record data for later use for tracking purposes.
Sensing the density of tissue can also be used to assist
the surgeon in identifying tissue. In open cardiac bypass
operations, the surgeons insert their hands inside the body
to palpate arteries. For cardiac bypass surgery, surgeons
select the bypass location by feeling where the fat and fatty
plaque is located in your arteries with their fingers. The
lack of tactile feedback in minimally invasive surgery,
prevents them from using this technique. The MEMS
devices have been developed for the palpation of tissue
using strain gauges (7), diaphragms (8), micropositioners
(9,10), and load cells (11) and have shown the ability to
measure blood pressure, pulse, different kinds of arterial
plaque, and distinguish between colon, bowel, stomach,
lung, spleen, and liver tissue.
Piezoelectric transducers can also be used to measure
density. Macroscale transducers are frequently used in
imaging applications to differentiate between tumors,
blood vessels, and different types of tissue. These transducers both emit and receive sound waves. By vibrating at a
high frequency sound waves are emitted in the direction of
the object of interest. The density of the impinged object
can then be measured based on the signal that is reflected
back by that object. Sound waves are reflected off the
interfaces between different types of tissue and returned
to the transducer. Present ultrasound systems tend to be
large and are not well suited for incorporation into microsurgical devices. The MEMS technology is well suited for
this application and many ultrasonic MEMS sensors have
been developed for imaging (1216).
Microelectromechanical systems ultrasound devices for
density measurements are shown in Fig. 11. They have
been shown to be able to detect the location of bone in tissue
and are being applied to atrial fibrillation surgeries. Atrial
fibrillation is what causes irregular heartbeats and leads to
one out of every six strokes. Drugs can be used to treat this
condition, but have dangerous side effects including causing a switch from atrial fibrillation to the more dangerous
ventricle fibrillation. Pacemakers and other electrical control devices can be used, but they do not always work for all
patients. The most effective treatment is the surgical
Figure 11. Ultrasound transducers next to a dime. (Courtesy of

Verimetra, Inc.)
530
MICROSURGERY
MAZE procedure, but it is an incredibly invasive treatment. The patient is put on a heart lung machine and then
their heart is stopped. Next, the surgeon takes a scalpel
and actually cuts all the way through the heart making
lesions, which physically separate the heart muscle. These
lesions break the electrical connections in the heart. The
heart is then sewn back together. Recently, there have
been a variety of different methods used to address the
problem. Instead of physically cutting all the way through
the heart with a scalpel, surgeons are using radio frequency, microwave, cryo, and laser energy to create transmural lesions. Transmurality means that the lesions go all
the way through the tissue, breaking the hearts electrical
connections. One of the problems surgeons encounter is to
know how deep the ablation is or if it is transmural. If the
lesions are not completely transmural or just partially
transmural then the undesirable electrical signals may
still be transmitted to the heart muscle. The MEMS ultrasound technology or micromachined electrodes can be used
to measure the transmurality.
Temperature can be used to detect if a surgical device is
close to a blood vessel, or if the surgical tool is in a diseased
or infected area. Temperature can also be used to monitor
the usage of a surgical device, by monitoring the time at
which the device is at body temperature. Usage is just one
of many areas where Auto-ID technologies will benefit
microsurgery (17). They can be used to make sure that
only the correct surgical tool is used for a procedure and if
that tool has been properly sterilized. Keeping track of how
many times, how long, and what was done with a surgical
tool will improve the reliability and effectiveness of surgical procedures and will greatly impact the entire medical
industry.
TRACKING SYSTEMS
Traditionally, a surgeon uses an endoscope in minimally
invasive surgery to determine where the surgical instruments are located in the patient. The view the surgeon has
of the surgical area is not ideal and the position of surgical
instruments outside of the camera view is not known.
Ideally, the surgeon would like to know the position and
orientation of each of his instruments. Computer-aided
surgery has enabled the surgeon to overlay magnetic resonance imaging (MRI) or computed artial tomography
(CAT) scan images of the patient with position and orientation data taken during surgery to create 3D models that
the surgeon can use to better visualize the surgical procedure. Computers can be used to simulate the procedure
beforehand allowing the surgeon to practice difficult operations ahead of time.
Current technology in this area is predominately optical
in nature. Markers are placed on the ends of the surgical
instruments that are located outside of the body, as well as
on specific locations on the patients body. A computer
registers the location of the surgical tools with the reference markers on the patient so that images of the patients
body can be aligned with the surgical tools. This is done
through the use of visible and infrared (IR) cameras. The
tips of the surgical tools that are located inside of the body
are then extrapolated. The markers must not interfere

with the surgery in any way, and therefore should be as
small and lightweight as possible. While these systems are
wireless and do not have cords that can get tangled on the
surgeon or on the surgical tools, there must be an unobstructed path from the markers to the camera systems. The
surgeon must be careful not to block the markers themself
or with other surgical instruments. Precision is compromised because the location of the surgical tips is extrapolated and does not take into account bending of the
surgical tools. Markers on the outside of the body do not
take into account compression of the tissue.
MEMS based ultrasound tracking systems have been
developed to address these issues (18). Constellations of
ultrasound sensors can also be placed on the surgical tools
themselves to determine position thereby eliminating
errors from extrapolation. Reference markers can now be
placed inside of the body, closer to the surgical area so that
they are less affected by compression and movement of the
patient.
Position and orientation can also be estimated using
accelerometers and gyroscopes. The signal outputs can be
integrated to determine or predict the distance traveled by
a surgical tool. Conventional MEMS accelerometers have
accuracies in the milligram range, which are not sufficient
for measuring accurately the relatively small displacements made during surgery (19). More accurate inertial
sensors need to be developed before they can be integrated
into surgical tools.
Magnetic field sensors can also be used to determine
position and orientation of surgical tools (20,21). A three
axis magnetoeffect sensor has been developed for determining the location of catheters. Currently this is done by
continually taking X rays of the patient. However X rays
only provide a 2D snapshot, a 3D image would be more
helpful for navigation.
EYE SURGERY
The leading cause of vision loss in adults > 60 are cataracts. The word cataract comes from the Greek meaning
waterfall and was originally thought to be caused by
opaque material flowing, like a waterfall, into the eye.
The condition is actually caused by the clouding of the
eyes intraocular lense. In the eye, light passes through
the lens that focuses it onto the retina. The retina converts
the light into electrical signals that are then interpreted
by the brain to give us our vision. The lens is a hard
crystalline material made mostly of water and protein.
The protein is aligned in such a way to allow light to pass
through and focus on the retina. When proteins in the lens
clump together, the lens begins to cloud and block light
from being focused on the retina. This causes vision to
become dull and blurry, which is commonly referred to as a
cataract.
Much like other surgery in other parts of the body,
cataract surgery has followed down the minimally invasive
path for the same benefits. Cataract surgery is one of the
earliest known surgical procedures. The earliest evidence
is the written Sanskrit writings of the Hindu surgeon
MICROSURGERY
Susrata dating from the fifth century BC. He practiced a

type of cataract surgery known as couching or reclination,
in which a sharp instrument was inserted into eye and
pushed the clouded lens out of the way. This displacement
of the lens enabled the patient to see better. Although
vision was still blurred without corrective lenses, many
famous people underwent this procedure including the
artists Michelangelo, Rembrandt, and Renoir. Couching
was still performed until the mid-twentieth century in
Africa and Asia.
In 1748, Jacques Daviel of Paris introduced extracapsular surgery where the lens was removed from the eye.
Later, very thick pairs of glasses were used to focus the
light onto the retina and restore sight, but the glasses were
cumbersome and caused excessive magnification and distortion.
By 1949, Dr. Harold Ridley of England, used PMMA as
the first intraocular lens. He discovered that PMMA was
biocompatible with the eye while treating WWII fighter
pilots whose eyes were damaged by shattering plastic from
their windshields. In the 1960s and 1970s, extracapsular
surgery became the preferred treatment. A large incision
(1012 mm) was made in the eye to remove and replace the
lense. This procedure minimized problems with image size,
side vision, and depth perception, but the large incisions
required longer hospitalization, recovery time, and
stitches.
Today, cataracts are removed with a procedure called
phacoemulsication with 1.5 million operations performed
yearly. A hollow ultrasonically driven titanium needle is
inserted into the anterior chamber of the eye. Ultrasonic
energy is then used to liquefy the hard lens and it is then
aspirated out of the eye. A foldable lens made of acrylic or
silicone is inserted through a (13 mm hole) as a replacement. Since the incision size has been reduced compared to
conventional extracapsular surgery, hospitalization, general anesthesia, sutures and bandages have all been eliminated. The reduction in incision size has also reduced the
risk of infection and postoperative refractions.
During the procedure the surgeon cannot see directly
under the needle as the lens is broken up and aspirated. A
thin clear membrane or capsule surrounds the lense. The
posterior capsule tissue underneath the lens is very delicate and easily cut compared the crystalline lens. To prevent the soft underlying tissue from damage requires a
skilled surgeon whom has performed many procedures. If
the posterior capsule is ruptured it can lead to glaucoma,
infection, or blindness. As the size of the incision has
decreased, heat damage to surrounding tissue from the
ultrasonic tip has increased that can alter the characteristics of the tissue and change its appearance. In addition
positive intraocular pressure must be maintained by balancing the flow of infusion fluid at positive pressure and
the aspirated cataract lens fragments. If pressure is not
maintained the anterior chamber can collapse. Pressure is
currently maintained by sensors located many feet from
the surgical area. This distance creates delays in the feedback loop, which can cause dangerous pressure fluctuations leading to damage to the iris and cornea.
Recently, micromachined silicon ultrasonic surgical
tools for phacoemulsifiaction have been developed by Lals
531
Figure 12. Ultrasonic phacoemulsification tool. (Courtesy of Jeff

Miller/University of Wisconsin-Madison.)
research group (22,23) (Fig. 12). Piezoelectric material is

attached to a micromachined silicon needle. The needle has
a fluid channel for aspiration as well as a horn for amplifying the ultrasonic displacement. These silicon devices are
able to generate higher stroke velocities and lower heat
generation than their conventional titanium counterparts.
High levels of integration has been achieved by integrating
pressure and flow sensors directly on the needle for maintaining intraocular pressure, reducing delays and making
the phacoemulsification procedure safer.
To prevent damage to the posterior capsule a piezoelectric sensor has been integrated into a phacoemulsification hand piece and undergone clinical trials (24). The
device can determine tissue hardness by measuring the
impressed loading on the needle tip or by monitoring
the resonant frequency at which the ultrasonic system
oscillates. Both of these approaches have proven successful
in determining when a hard to soft tissue transition has
occurred during a phacoemulsification procedure. This
technology enables a surgeon to get real-time feedback
on the type of tissue he is cutting, and can be applied to
other types of surgical procedures such as tumor extraction
as well.
Insertion of a replacement lense requires precise movements by the surgeon. Serious postoperative vision problems may occur if the lens is inserted incorrectly and
needs to be removed. Precision piezoelectric micromotors
have been developed for intraocular delivery of replacement lenses after cataract removal (25). These inchworm
actuators use a glider and clamp arrangement to generate
large forces over small displacements. An electrostatic
clamp made of an oxide dielectric layer sandwiched
532
MICROSURGERY
between two silicon wafers layer locks the micromotor in

place while a PZT actuator generates the force. The inertia
of a mass is used to move the clamp. Forces of 3.0 N and
step sizes of 100 nm to 10 mm have been reported.
In eye surgery there are many times when precision
cutting is required. Highly sharpened steel, ceramic, or
diamond scalpel blades are used, but are expensive to
produce costing up to $1000 a piece. Disposable silicon
micromachined scalpels are an attractive alternative. They
can be batch fabricated to reduce cost and sharpened to an
atomic edge along their crystal planes. They are already
being used in Russia at the Fyodorov Eye Care Center in
nonpenetrating deep sclerectomy operations for the treatment of glaucoma (26). BD (Bekton, Dikenson, and Company) is producing Atomic Edge silicon blades for cataract
surgery (27). Soon they will be used for eye operations and
eventually migrate to procedures on other parts of the body.
Smaller incisions made by sharper blades result in less
bleeding and tearing. An added advantage of silicon blades
is that sensors and electronics can be directly fabricated on
them during fabrication. Integrating cauterizing electrodes
on the blade itself will prevents the patient from bleeding as
well as let the surgeon more clearly see the surgical area.
CATHETERS/GUIDEWIRES/STENTS
Cardiac catheterizations can be referred to as a noninvasive surgical procedure. Specialized tubes or catheters are
threaded up through blood vessels in the groin, arm, or
neck to an area of the body which needs treatment. The
problem is then treated from the inside of the blood vessel.
The advantage of these approaches is that the procedures
require very small incisions, hospital stays are usually one
night or less and the discomfort and recovery times afterwards are minimal. For patients with more complicated
ailments, catheter treatments can be used in combination
with minimally invasive or open surgery to give the best
possible results at the lowest possible risk. Catheters,
guidewires, and stents currently represent the most widespread use of MEMS technology in surgery.
Diagnostic catheterizations, which are used to measure
pressure in different parts of the body, take blood samples,
and to perform detailed angiograms of the heart, can be
performed by injecting X-ray dye through the catheters.
The MEMS pressure sensors are now commonly found on
catheter devices and are the most mature MEMS technology in this area. Even smaller designs are being sold for
placement on guidewires, such as those made by Silex
Microsystems, shown next to a 30 gauge needle (Fig. 13).
Each sensor is but 100 mm thick, 150 mm wide, and 1300 mm
long (28). MEMS ultrasound sensors are also starting to
be used for both forward looking (16) and side looking
intravascular imaging (12,13).
To provide the doctor with more information to make
better decisions, additional MEMS sensors are needed to
gather additional data for the diagnosis, monitoring of
procedures, as well as for checking results of completed
operations. Many other types of MEMS sensors are being
researched to measure blood flows, pressures, temperatures, oxygen content, and chemical concentrations for
placement on diagnostic catheters (29,30).
Figure 13. MEMS pressure sensors next to a 30-gauge needle

(Courtesy of Silex Microsystems, Jarfalla, Sweden.)
Heart disease continues to be the leading cause of death

in the United States. Interventional catheterization is an
increasingly more common way to treat blood vessels which
have become occluded (blocked) or stenotic (narrowed) by
calcified artherosclerotic plaque deposits. Blood vessels
that have become occluded or stenotic may interrupt blood
flow, which supplies oxygen and cause heart attacks or
strokes. Occluded or stenotic blood vessels may be treated
with a number of medical procedures including angioplasty
and atherectomy. Angioplasty techniques, such as percutaneous transluminal coronary angioplasty (PTCA), also
known as balloon angioplasty are relatively noninvasive
methods of treating restrictions in blood vessels. A balloon
catheter is advanced over a guidewire until the balloon is
positioned in the restriction of a diseased blood vessel. The
balloon is then inflated compressing the atherosclerotic
plaque. Frequently, the wall of the vessel is weakened
after inflation and a metal stent in is expanded and
deployed against the plaque. The stent helps keep the
vessel open. During an atherectomy procedure, the stenotic
lesion is mechanically cut or abraded away from the blood
vessel wall using an atherectomy catheter.
Microelectrochemical systems pressure sensors can be
used to measure the pressure in the balloons during inflation, to make sure damage due to over inflation is minimized. The MEMS temperature sensors can be integrated
on catheters and guidewires to determine the location of
inflamed plaque. The inflammation causes artery walls in
the damaged area to have an increased temperature up to
38C higher than healthy tissue. Approximately 2050% of
all patients undergoing these therapeutic procedures to
clear blocked coronary arteries will suffer restenosis
(reblockage) within 6 months of the initial procedure. Drug
coated stents have significantly lowered these rates and
have been approved for use in Europe for a few years. They
are expected to be approved in the United States later this
year. The MEMS laser micromachining technology is used
in the fabrication of conventional stents and drug coated
stents (31). Stainless steel micromachining technology has
also been developed at the University of Virginia for piercing structure drug delivery/gene therapy stents for the
treatment of restenosis (32). There is potentially a large
opportunity for MEMS in embedding sensors into stents to
MICROSURGERY
533
create a smart stents, which would be able to alert doctors

when restenosis occurs or other problems occur (33). The
MEMS rotary cutting devices have been fabricated for
atherectomy procedures (34), but are not widely used
because cut up plaque particles can flow downstream
and cause strokes.
FETAL SURGERY
Fetal surgical techniques were first pioneered at the University of California San Francisco (UCSF) in the 1980s to
operate on babies while they were still in the womb. The
success rate of treating certain birth defects is higher the
earlier they are addressed in fetal development. Initially
open surgical techniques were used with an incision through
the mothers abdomen to allow direct access for the surgeons hands. After the surgery was complete, the womb was
sutured and the mother delivered the baby weeks or months
later. In 1981, the first fetal urinary tract obstruction procedure was performed at UCSF. Lately, minimally invasive
surgical and robotic surgical techniques have been used that
have reduced the risk of complications and premature labor.
Other fetal procedures have also been developed to treat
cojoined twins, hernias, spina bifida, tumors, and heart
defects. One area of interest is in fetal heart.
The development of the cardiovascular system in a
developing fetus is typically completed by the twelfth week
of gestation. At this stage primary heart defects are small
and if repaired will prevent defects from becoming more
severe as the heart changes to adapt to the normalization
blood flows and pressures in the later periods of gestation.
Fetal heart surgery can be used to treat hypoplastic
heart syndrome (HHS), which was often considered fatal.
This syndrome causes the hearts left side to fail to grow
properly and is caused by an obstruction which restricts
blood flow. The heart requires the mechanical stress of
blood flow to grow properly, and thus fails to develop
normally. Today, three open surgeries are required to allow
the remaining single (right) ventricle to support both the
pulmonary and systemic circulations. The long-term survival for these patients into adulthood is significantly
< 50%. Preserving both ventricles would result in the best
chances of long-term survival.
An interventional catheter can be used to treat HHS in a
minimally invasive manner. The balloon catheter must
first penetrate in through the mothers abdominal wall,
the placenta and then the fetuss chest into its heart. It
must then locate the fetuss tiny heart and expand to open
the blockage. Afterward, the surgeon needs to know
whether the operation has succeeded enough for the baby
to develop normally. Verimetra, Inc., Carnegie Mellon
University, and Childrens Hospital of Pittsburgh have
embedded a MEMS flow sensor and a series of micromachined bar codes on the tip of a balloon catheter. The bar
codes enable the catheter to be visualized with ultrasound
allowing surgeons to know its exact position. The flow
sensor is a thermistor. As blood flow increases the temperature measured by the thermistor decreases and in this
manner blood flow changes as the catheter progresses
through the hearts vessels can be monitored. This allows
for the measurement of blood flow at or near a constriction
Figure 14. Highly integrated microsurgical probe. (Courtesy of

Prof. Dr. M. Schurr, novineon Healthcare Technology Partners
GmbH and Prof. G. Buess, Universita tsklinikums Tbingen.)
and then again after the procedure to open the constriction

has been performed.
FUTURE SYSTEMS
In 1959, Richard Feynman gave his famous talk Theres
Plenty of Room at the Bottom (35). In it, he talked about
being able to put a surgeon in a blood vessel which would be
able to look around ones heart. Initially, new microsurgical
tools will focus on measuring or detecting a specific parameter be it pressure, blood flow, velocity, temperature, and
so on. The MEMS sensor systems will continue to be refined
and improved leading to the integration of multiple sensor
systems on surgical tool followed by tools with multiple
functions. Multifunction surgical tools will reduce the
number of tool insertions and removals reducing patient
risks. Eventually, this will lead to highly integrated probes,
which will do everything a surgeon needs, such as the
concept shown in Fig. 14 (36). These tools will fit through
a standard 5-mm port and have built in 3D cameras for
visualization, biopsy samplers with microfluidic processing
capability to do tissue analysis, ultrasound transducers,
and tactile sensors for feedback to the surgeon.
BIBLIOGRAPHY
1. Diamiano R. Next up: Surgery by remote control. NY Times,
Apr. 4, 2000; pp. D1.
2. Blazer E, Koh W, Yon E. A miniature digital pressure transducer. Proceedings of the 24th Annual Conference on Engineering Medicine and Biology. 1971. pp 211.
3. Peirs J, et al. A microturbine made by micro-electro-discharge
machining. Proc 16th Eur Conf Solid-State Transducers
2002; 790793.
4. Pique A, et al. Laser direct-write of embedded electronic
components and circuits. Presented at Photon Processing
534
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
MICROSURGERY
in Microelectronics and Photonics IV, Jan 2427 2005, San
Jose, (CA); 2005.
Go tzen R. Growing up, additive processes in MEMS fabrication and packaging. Presented at Machines and Processes
for Micro-scale and Meso-scale Fabrication, Metrology and
Assembly, Gainesville, FL; 2003.
Rebello KJ, Lebouitz K, Migliuolo M. MEMS tactile sensors
for surgical instruments. MRS Symp. Proc.: Biomicroelectromech. Systems (BioMEMS). 2003;773:5560.
Menciassi A, et al. Force feedback-based microinstrument for
measuring tissue properties and pulse in microsurgery. Presented at 2001 IEEE International Conference on Robotics
and Automation (ICRA), May 2126 2001, Seoul; 2001.
Rebello KJ, et al. MEMS based technology for endoscopic
assessment of coronary arterial hardness. Presented at the
5th Annual NewEra Cardiac Care Conference, Dana Point
(CA); 2002.
Bicchi A, et al. Sensorized minimally invasive surgery tool for
detecting tissutal elastic properties. Presented at Proceedings of
the 1996 13th IEEE International Conference on Robotics and
Automation. Pt. 1 (of 4), Apr. 2228 1996, Minneapolis, (MN); 1996.
Rosen J, Hannaford B, MacFarlane MP, Sinanan MN. Force
controlled and teleoperated endoscopic grasper for minimally
invasive surgeryexperimental performance evaluation.
IEEE Trans Biomed Eng 1999;46:12121221.
Scilingo EP, Bicchi A, De Rossi D, Iacconi P. Haptic display
able to replicate the rheological behaviour of surgical tissues.
Presented at Proceedings of the 1998 20th Annual International Conference of the IEEE Engineering in Medicine and
Biology Society. Pt. 4 (of 6), Oct. 29-Nov. 1 1998, Hong Kong,
China; 1998.
Fleischman A. et al. Miniature high frequency focused ultrasonic transducers for minimally invasive imaging procedures. Sensors Actuators, A: Phys 2003;103:7682.
Zara JM, Bobbio SM, Goodwin-Johansson S, Smith SW.
Intracardiac ultrasound scanner using a micromachine
(MEMS) actuator. IEEE Trans Ultrasonics, Ferroelectrics,
and Frequency Control 2000;47:984993.
Daft C, et al. Microfabricated ultrasonic transducers monolithically integrated with high voltage electronics. Presented
at 2004 IEEE Ultrasonics Symposium, Aug. 2327 2004,
Montreal, Queeue, Canada; 2004.
Chang JK, et al. Development of endovascular microtools. J
Micromech Microeng 2002;12:824831.
Degertekin FL, Guldiken RO, Karaman M. Micromachined
capacitive transducer arrays for intravascular ultrasound.
Presented at MOEMS Display and Imaging Systems III, Jan.
2425 2005, San Jose, (CA); 2005.
Brock D. Smart medicine: The application of Auto-ID technology
to healthcare. Auto-ID Center, MIT, Cambridge, (MA); 2002.
Tatar F, Mollinger JR, Bastemeijer J, Bossche A. Time of
flight technique used for measuring position and orientation
of laparoscopic surgery tools. Presented at Sensors, 2004.
Proceedings of IEEE; 2004.
Fang C-M, Lee S-C. A research of robotic surgery technique
by the use of MEMS accelerometer. Presented at Engineering
in Medicine and Biology, 2002. 24th Annual Conference and
the Annual Fall Meeting of the Biomedical Engineering
Society. EMBS/BMES Conference; 2002. Proceedings of the
Second Joint, 2002.
Tanase D, et al. 3D position and orientation measurements
with a magnetic sensor for use in vascular interventions.
Presented at Biomedical Engineering, 2003. IEEE EMBS
Asian-Pacific Conference on; 2003.
Totsu K, Haga Y, Esashi M. Three-axis magneto-impedance
effect sensor system for detecting position and orientation of
catheter tip. Sensors Actuators, A: Phys 2004;111:304309.
22. Chen X, Lal A. Integrated pressure and flow sensor in siliconbased ultrasonic surgical actuator. Presented at Ultrasonics
Symposium, 2001 IEEE; 2001.
23. Son I-S, Lal A. Feedback control of high-intensity silicon
ultrasonic surgical actuator. Presented at Solid-State Sensors, Actuators and Microsystems, 2005. Digest of Technical
Papers. TRANSDUCERS 05. The 13th International Conference on; 2005.
24. Polla DL, et al. Microdevices in medicine. Annu Rev Biomed
Eng 2000;2:55176.
25. Polla D, et al. Precision micromotor for surgery. Presented at
Microtechnologies in Medicine and Biology, 1st Annual International, Conference On. 2000; 2000.
26. Kozlova TV, Shaposhnikova NF, Scobeleva VB, Sokolovskaya
TV. Non-penetrating deep sclerectomy: Evolution of the
method and prospects for development (review). Ophthalmosurgery 2000;3:3954.
27. Angunawela R, Von Mohrenfels CW, Marshall J. A new age of
cataract surgery. Cataract & Refractive Surgery Today I
2005; 3638.
28. Kalvesten JE, Smith L, Tenerz L, Stemme G. First surface
micromachined pressure sensor for cardiovascular pressure
measurements. Presented at Proceedings of the 1998 IEEE
11th Annual International Workshop on Micro Electro Mechanical Systems, Jan. 2529 1998, Heidelberg, Ger; 1998.
29. Tanase D, Goosen JFL, Trimp PJ, French PJ. Multi-parameter sensor system with intravascular navigation for catheter/guide wire application. Presented at Transducers01
Eurosensors XV, Jun. 1014 2001; Munich, 2002.
30. Haga Y, Esashi M. Biomedical microsystems for minimally
invasive diagnosis and treatment. Proc IEEE Biomed App
Mems Microfluidics 2004;92:98114.
31. Kathuria YP. An overview on laser microfabrication of biocompatible metallic stent for medical therapy. Presented at
Laser-Assisted Micro- and Nanotechnologies 2003, Jun. 29
Jul. 3 2003, St. Petersburg, Russian Federation; 2004.
32. Reed ML. Micromechanical systems for intravascular drug
and gene delivery. Presented at BioMEMS 2002 Conference,
Boston; 2002.
33. Goldschmidt-Clermont P, Kandzari D, Khouri S, Ferrari M.
Nanotechnology needs for cardiovascular sciences. Biomed
Microdevices 2001;3:8388.
34. Ruzzu A, et al. A cutter with rotational-speed dependent diameter for interventional catheter systems. Presented at Micro
Electro Mechanical Systems, 1998. MEMS 98. Proceedings.,
The Eleventh Annual International Workshop on, 1998.
35. Feynman RP. Theres plenty of room at the bottom. J Microelectromech Systs 1992;1:6066.
36. Schurr MO, Heyn S-P, Ment W, Buess G. Endosystems
Future perspectives for endoluminal therapy. Minimally
Invasive Ther Allied Technol 1998;7:3742.
Further Reading
Taylor RH, Lavallee S, Burdea GC, Mosges R. Computer Integrated Surgery: Technology and Clinical Application. Cambridge,
(MA): MIT Press; 1996.
Zenati M. Robotic heart surgery. Cardiol Rev 2001;9(5):18.
Davies B. A review of robotics in surgery. Proc Inst Mech Eng
2000;214:129140.
Madou M. Fundamentals of Microfabrication: The Science of
Miniturization. 2nd ed. Boca Raton, (FL): CRC Press; 2002.
Kovacs GTA. Micromachined Transducers Sourcebook. Boston,
MA: McGraw-Hill; 2002.
See also ENDOSCOPES; FIBER OPTICS IN MEDICINE; INTRAUTERINE SURGICAL
TECHNIQUES; NANOPARTICLES; RADIOSURGERY, STEREOTACTIC.
MINIMALLY INVASIVE SURGICAL TECHNOLOGY
MINIMALLY INVASIVE SURGICAL

TECHNOLOGY
JAY R. GOLDBERG
Marquette University
535
procedures are performed with devices located outside the

body, and thus are noninvasive. The MIS procedures are an
alternative to open surgery. The benefits of MIS include
reduced patient trauma, postoperative recovery time, and
healthcare costs.
INTRODUCTION
INSTRUMENTATION FOR MINIMALLY INVASIVE SURGERY
Most surgical procedures involve the invasion and disruption of body tissues and structures by surgical instrumentation and/or implantable medical devices, resulting in
trauma to the patient. Diagnostic imaging procedures,
such as magnetic resonance imaging (MRI), computed
tomography (CT), X ray, positron emission tomography
(PET), and ultrasound do not require disruption of body
tissues, and are thus considered to be noninvasive. Extra
corporeal shockwave lithotripsy (ESWL) used to disintegrate kidney stones is an example of a noninvasive therapeutic procedure. As shown in Fig. 1, it focuses acoustic
energy, generated outside of the patients body, on kidney
stones. No trauma to the patient occurs during this
procedure.
Open heart coronary artery bypass and organ transplantation surgery are examples of highly invasive surgery. These procedures require a high level of invasion and
disruption of body tissues and structures. Bones, muscle
tissue, and blood vessels are cut, and tissue from other
parts of the body may be grafted, resulting in a high level of
surgical trauma to the patient.
Minimally invasive surgical procedures are less traumatic than corresponding conventional surgical procedures. The use of small instruments placed through
intact natural channels, such as the esophagus, urethra,
and rectum, is less invasive than conventional open surgical approaches requiring large incisions, significant loss of
blood, and trauma to tissues. The use of small instruments,
such as biopsy guns and angioplasty balloons placed
through small incisions, results in minor trauma to the
patient, but is much less invasive than open surgical
procedures used to accomplish the same goals. Procedures
using small instruments through intact natural channels
or small incisions are classified as minimally invasive.
A minimally invasive surgical procedure can be defined
as surgery that produces less patient trauma and disruption of body tissues than its conventional open surgical
counterpart. For example, conventional appendectomies
require a 4 cm long incision made through layers of skin,
muscle, and other tissues to gain access to the appendix.
Once the appendix is removed, the layers of the wound are
sutured together to allow them to heal. This invasive
procedure requires a significant level of invasion and disruption of tissues and other structures. The minimally
invasive appendectomy is performed with small surgical
instruments placed into small incisions in the patients
abdomen. Once the appendix is removed, the small incision
is closed with sutures. This procedure results in much less
trauma to the patient than the open surgical approach.
Minimally invasive surgery (MIS) is performed with
small devices inserted through intact natural orifices or
channels, or small incisions used to create an orifice. Some
Many MIS procedures involve flexible or rigid fiber optic

endoscopes for imaging surgical sites and delivering
instrumentation for diagnostic or therapeutic applications.
The term endoscope is a generic term used to describe
tubular fiber optic devices placed into the body to allow
visualization of anatomical structures.
Endoscopes consist of a fiber optic light guide, high
intensity light source, coherent fiber optic bundle, steering
mechanism, and various working channels for insertion of
endoscopic instrumentation and infusion of irrigation
fluids or insufflating gases. Glass fibers comprise the fiber
optic light guide and coherent bundle and are surrounded
by a flexible polyurethane sheath or rigid stainless steel
tube. Figure 2 shows the components of a typical endoscope. The light source provides light that is transmitted
through the light guide and projected onto the anatomical
site to create an image. The coherent fiber bundle transmits the image back to the focusing eyepiece and into the
surgeons eye. The surgeon can move the endoscope proximally (toward the patient) and distally (away from the
patient) by pushing and pulling the endoscope into and out
of the body, respectively. Steering is accomplished with a
handle attached to a cable that pulls and bends the tip of
the endoscope in the desired direction. The proximal (closest to the patient) end of the endoscope (shown in Fig. 3)
contains lumens for the fiber optic light guide and coherent
fiber bundle, and one or more working channels for passage
of instruments and irrigation fluid into and out of the
operative site. Some endoscopes contain video cameras
that display endoscopic images on a video monitor in the
operating room, as shown in Fig. 4.
Specialized endoscopes have different names depending
on what part of the body they are used to image. Table 1
lists the names of various endoscopes, the procedures for
which they are used, and the anatomical location where
they are used.
Endoscopic Procedures and Instrumentation
To perform an endoscopic procedure, the surgeon inserts a
sterilized endoscope into a natural channel or orifice, such
as the urethra, rectum, esophagus, bronchial tube, or nose.
If there is no natural channel to provide access to the
operative site (as in abdominal surgery), then the surgeon
will create one with a small incision, and may insert a
hollow trocar into the incision, as shown in Fig. 5. The
trocar is left in the wound to provide access to the abdominal cavity. The endoscope is inserted into the access site
(natural channel, incision, or trocar) and as it is advanced
toward the operative site the surgeon monitors the image
produced by the endoscope, either through the objective
eyepiece or video monitor (as shown in Fig. 4). Rigid
endoscopes are typically used when access is obtained
536
Figure 1. Schematic diagram of noninvasive

extra corporeal shockwave lithotripsy (ESWL)
procedure used to disintegrate kidney or biliary
stones. Acoustic waves generated from spark
plugs are focused with ellipsoidal reflectors on
the kidney stone.
through an incision. Flexible endoscopes are used when

access is obtained through natural channels (2). Once in
position, the surgeon can then manipulate the endoscope to
inspect tissues and anatomical structures (Table 1).
If a procedure other than visual inspection is required,
the surgeon has a variety of instruments available to grasp,
cut, remove, suture, and cauterize tissues, and remove debris
through the endoscope. Forceps (Fig. 6) and graspers are
used to grasp tissue and other objects such as kidney
stones. Scalpels and scissors (Fig. 7) are used for cutting

tissue. Suturing devices are used to repair internal wounds
resulting from endoscopic procedures such as appendectomies, cholecystectomies (gallbladder removal), and
arthroscopies. Morcellators are used to reduce the size
and change the shape of a mass of tissue, such as a
gallbladder, allowing it to be removed through a small
incision. Electrohydraulic lithotriptor (EHL) and laser
probes are used to disintegrate objects, such as kidney
Coherent fiber bundle

Working channel
Proximal
end of
endoscope
Flexible fiber optic shaft
Coherent fiber
bundle
Fiber optic
light
guide
Eyepiece with lens

(or video camera)
Surgeons eye
Anatomical site
Fiber optic
light
guide
Light
source
Figure 2. Components of a typical endoscope.
537
Figure 3. Proximal ends of typical endoscopes. Endoscopic

instruments (left:scissors, right:forceps) are shown placed
through working channels. Two other channels contain the fiber
optic light guide and coherent fiber bundle, respectively. A fourth
channel can be used for instrumentation or delivery of irrigation
fluids to operative site (1). (Reprinted from Endoscopic Surgery,
Ball, K., page 62, Copyright 1997, with permission from Elsevier.)
stones, with acoustic and laser energy, respectively, allowing the debris to be passed through the ureter. Stone
baskets are used to trap kidney or ureteral stones for
endoscopic removal, as shown in Fig. 8. These instruments
are placed through the working channel of an endoscope or
trocar and are operated by manipulating handles and
controls located outside the patient.
Figure 5. Trocar inserted into abdominal incision to provide

access to abdominal cavity. Safety shield covers sharp end of
trocar (circled) upon insertion to prevent damage to abdominal
organs. (1) (Courtesy Ethicon Endo-Surgery Inc., Cincinnati, OH.)
Endoscopes may contain multiple working channels

that allow for irrigation of fluids to irrigate and flush out
clots and other surgical debris. The presence of the fluid
improves visibility through the endoscope by keeping the
visual field clear and the lens clean. Laparoscopic surgery
requires insufflation of CO2 or N2O through the working
channel or trocar and into the abdominal cavity to cause
the cavity to expand, separating organs, and enlarging the
operative site and visual field.
There are typically not enough working channels in a
laparoscope to accommodate all of the instruments needed
for a particular procedure. In these cases, multiple access
sites are created with additional trocars. Typically, a camera is placed through one trocar and used to visualize the
work being performed with instruments placed through
other trocars.
When an endoscopic procedure is completed, the endoscope and instrumentation are removed from the patient
via the natural channel or incision. When a laparoscopic
procedure is completed, the laparoscope, camera, instruments, and trocars are removed from the patient. The
wounds created by the trocars are sutured and the patient
begins the recovery process. Although some of the CO2
from insufflation may escape the abdominal cavity when
all instrumentation is removed, some will be absorbed by
body tissues and eliminated via respiration. Patients typically recover and are discharged from the hospital within
2448 h.
Non-Endoscopic Procedures and Instrumentation
Figure 4. Surgeon viewing laparoscopic images of gallbladder on

video monitor in operating room. Note use of three trocars in
patients abdomen; one each for video camera and two
instruments. (Courtesy ACMI, Southborough, MA.)
Not all minimally invasive procedures require endoscopic

devices for imaging and placement of instruments. Some
MIS procedures (listed in Table 2), such as stereotactic
radiosurgery, use lasers or gamma radiation in place of
scalpels. Others use small catheters to deliver medication,
devices, or energy to specific anatomical locations. Balloons
and stents, delivered via catheters, are used to access and
dilate constricted vessels and maintain patency. Catheter
mounted electrodes are used to deliver thermal, micro-
538
Table 1. Names of Endoscopes, MIS Procedures with Which they are Used, and Anatomical Location Where they are Used
Medical Specialty
Type of Endoscope Used
MIS Procedures
Anatomical Location
Urology
Cystoscope
Ureteroscope
Nephroscope
Urethra, bladder
ureter, kidney
Gastroenterology
Gastroscope
Colonoscope
Sigmoidoscope
Laparoscope
Cystoscopy,
Transurethral resection of the
prostate (TURP)
Ureteroscopy, stone removal
Nephroscopy, stone removal
Gastroscopy, gastric bypass
Colonoscopy, Sigmoidoscopy
General surgery
Orthopedics
Ob/Gyn
Ear, nose, and
throat
Arthroscope
Hysterescope
Laryngoscope,
Bronchoscope
Rhinoscope
wave, or radio frequency energy to selectively destroy

tissue in ablation procedures used to treat cardiac arrhythmias (3), benign prostatic hypertrophy (BPH) (4), and other
conditions. Many of these devices are guided through the
body with the help of imaging or surgical navigation equipment.
Image Guided SurgerySurgical Navigation
Image guided surgery (IGS) allows surgeons to perform
minimally invasive procedures by guiding the advancement of instrumentation through the patients body with
increased accuracy and better clinical outcomes. Preoperatively, an IGS system is used to produce computerized
anatomical maps of the surgical site from MRI or CT
images of the patient. These maps are then used to plan
the safest, least invasive path to the site. During an image
guided procedure, the IGS system provides surgeons with a
three dimensional image showing the location of instruments relative to the patients anatomical structures. It
tracks the movement of surgical instruments in the body,
correlates these movements with the patients preoperative images, and displays the location of the instruments on
a monitor in the operating room. This feedback helps the
Laparoscopy, hernia repair,

appendectomy, cholecystectomy
(gallbladder removal)
Arthroscopy
Tubal ligation, hysterectomy
Laryngoscopy, bronchoscopy
Rhinoscopy, sinuscopy
Stomach, colon
Sigmoid colon
Abdomen
Knee and other joints

Female reproductive tract
Larynx, bronchus, nose,
sinus cavities
surgeon safely and accurately guide instruments to the

surgical site, reducing the risk of damaging healthy tissue
(4,5).
An IGS system includes a computer workstation, localization system, display monitor, and specialized surgical
instruments capable of being tracked by the system. Image
processing and surgical planning software are also used.
Tracking of instruments is accomplished through optical,
electromagnetic, or mechanical means. Optical tracking
systems use a camera mounted to view the surgical field
and optical sensors attached to surgical instruments.
These systems required line of sight between the camera
and sensor to function properly. Electromagnetic tracking
systems include a transmitter located close to the surgical
site, and receivers attached to surgical instruments. Line of
sight is not an issue with these systems, however, nearby
metallic objects may produce interference to signals used to
track instruments (4).
To ensure that the patients anatomical features (and
location of instruments) are accurately displayed by the
IGS system, actual anatomical locations must be registered
to the preoperative images. This can be accomplished by
touching a probe to a marker on the patients body and then
assigning the location of this marker to its corresponding
point in preoperative images (4).
Figure 6. Endoscopic forceps. (Courtesy ACMI, Southborough,

MA.)
Figure 7. Endoscopic scissors. (Courtesy ACMI, Southborough,

MA.)
Figure 8. A stone basket used to trap and remove a large ureteral

stone. (Courtesy ACMI, Southborough, MA.)
539
Table 2. Examples of Non-Endoscopic Minimally Invasive Surgical Procedures

Medical Specialty
Non-Endoscopic Minimally Invasive Surgical Procedure
Cardiovascular surgery
Minimally invasive direct coronary artery bypass (MIDCAB)

Percutaneous transluminal coronary angioplasty (PTCA)
Coronary stenting
Radio frequency cardiac ablation
Laser angioplasty
Microwave catheter ablation for arrhythmias
Chemical ablation for ventricular tachycardia
Laser photorefractive keratotomy (PRK)
Laser ablation of cataracts
Laser glaucoma surgery
Total joint arthroplasty (hips, knees, and others)
Stereotactic radiosurgery
Stereotactic radiotherapy
Laser ablation of brain tissue, tumor tissue
Clot removal
Aneurism repair
Cerebral arterial venous malformation repair
Transjugular hepatic portal systemic shunt creation
Ophthalmology
Orthopedics
Neurosurgery
Radiology
Most surgical instruments must be adapted for use in

image guided surgery by mounting sensors and other
devices to allow detection of the instruments position by
the IGS system. Some medical device manufacturers are
developing navigation-ready surgical instruments that
contain small reflective spheres to act as reference arrays
for cameras used in image guided surgery (5).
MINIMALLY INVASIVE SURGICAL PROCEDURES
This section contains a few examples of minimally invasive
surgical procedures used in urology, orthopedics, neurosurgery, and general and cardiovascular surgery.
Ureteroscopy
Flexible ureteroscopy is used to perform a variety of diagnostic and therapeutic urological procedures. It involves
entry into the body via intact natural channels (urethra
and ureter) and does not require an incision. Local anesthesia and sedation of the patient are required.
Ureteroscopy is commonly used to remove ureteral or
kidney stones. Initially, a cystoscope is inserted into the
urethra. Next, a guidewire is placed through a cystoscope
into the ureter, and advanced up into the kidney. While
maintaining the position of the guidewire, the cystoscope is
removed, and the distal end of the guidewire is placed into
a working channel at the proximal end of the ureteroscope.
The ureteroscope is then advanced along the guidewire
into the ureter or kidney, as shown in Fig. 9. Active
(controlled by the cable and handle) and passive deflection
of the shaft, along with rotation of the flexible ureteroscope
allows visual inspection of the renal calices as shown in
Fig. 10. Ureteroscopic instruments are then placed through
the working channel into the ureter or kidney. Figure 11
shows two devices used for ureteroscopic removal of ureteral stones. The stone grasper is used to physically grab the
stone (Fig. 11). If the stone is small enough to fit into the
working channel, it is pulled out of the patient through the
ureteroscope. Large stones that will not fit into the working
channel are pulled out with the ureteroscope. The laser

lithotripter (Fig. 11) disintegrates the stone into small
particles that can easily be passed naturally through the
ureter, bladder, and urethra. Collapsed stone baskets can
be placed alongside a stone and moved proximally and
distally as they are expanded, until the stone is trapped
in the basket (as shown in Fig. 8) and pulled out of the
urethra.
Laparoscopy
Laparoscopy is commonly used for removal of the gallbladder and appendix, hernia repair, and other abdominal
procedures. The basic steps involved in laparoscopy have
been previously described. The lack of natural channels
located in the abdomen requires the use of trocars to gain
access to the operative site. Laparoscopic procedures
require insufflation of gas to separate organs and expand
the visual field. This is controlled by a separate insufflator
that controls the pressure inside the abdomen produced by
the flow of insufflating gases (1).
Figure 9. A surgeon views images through a ureteroscope placed

through the urethra, bladder, ureter, and into the kidney.
(Courtesy ACMI, Southborough, MA.)
540
Figure 10. Flexibility and steerability of proximal end of

ureteroscope allow inspection of upper, middle, and lower renal
calices. (Courtesy ACMI, Southborough, MA.)
Total Joint Replacement

Total hip and knee replacements are typically performed
with large incisions to expose the joint, allowing for
complete visualization of and access to the joint and soft
tissues. New surgical approaches using smaller incisions
that result in less damage to muscles and other soft tissue
limit the view of the joint. To compensate for the limited
view, fluoroscopy and IGS systems are often used. Some
existing surgical instruments have been modified to enable
surgery through smaller incisions (6).
Traditional hip arthroplasty requires a 3046 cm long
incision, which is highly disruptive to muscles and surrounding tissues. Two different techniques can be used for
minimally invasive total hip arthroplasty. One technique
Figure 11. Devices used for endoscopic removal of ureteral

stones. (Courtesy ACMI, Southborough, MA.)
Figure 12. Acetabular component inserted with inserter through

small incision (top image). Fluoroscopic image of inserter and
acetabular component seated in acetabulum (bottom image).
Images such as these allow surgeons to ensure proper
placement of instruments and implantable components within
hip joint when small incisions prevent viewing of entire device
(6). (From MIS for the Hip and Knee: A Clinical Perspective, 2004,
pg. 20, Minimally Invasive Total Hip Arthroplasty: The Two
Incision Approach, Berger, R.A. and Hartzband, M.A., Fig. 2.11.
Copyright 2004, with kind permission of Springer Science and
Business Media.)
uses two 5 cm long incisions, one each for preparation and

insertion of the acetabular and femoral components,
respectively. The other technique involves one 810 cm
long incision. Modified retractors and elevators are used
to gain access and expose the joint. Fluoroscopy and IGS
systems are used to provide the surgeon with a view of
instruments and components (as shown in Fig. 12) and
assist in positioning of instruments designed to enable
accurate component alignment and placement. Minimally
invasive hip arthroplasty results in less disruption
of muscles and tissues, smaller and less noticeable scars,
less blood loss and pain, and fewer blood clots and dislocations (6).
Most minimally invasive knee arthroplasties performed
through smaller incisions involve a single compartment of
the knee. These procedures typically use existing unicompartmental knee implants for resurfacing the medial or
lateral femoral condyle and corresponding tibial plateau.
Existing instrumentation has been modified to obtain
access to the joint and enable accurate placement and

alignment of the unicompartmental knee components
through smaller incisions.
Minimally Invasive Direct Coronary Artery Bypass
Coronary arteries may become blocked due to fatty deposits
(plaque), blood clots, or other causes. This reduces blood
flow to cardiac muscle, depriving it of needed oxygen and
other nutrients. This condition can result in a myocardial
infarction (heart attack) that can damage cardiac muscle.
Traditional open heart coronary artery bypass graft
surgery has been used to restore normal blood flow to
cardiac muscle. This procedure involves grafting a new
vessel to points on both sides of the blocked coronary
artery, thereby bypassing the blockage and restoring normal blood flow. It requires open access to a still heart to
allow suturing of the graft to the blocked coronary artery. A
sternotomy and separation of the sternum is required to
provide access to the heart. The heart is stopped and the
patient is attached to a heartlung machine to maintain
circulation of oxygenated blood through the body during
the surgical procedure. This procedure is highly invasive
541
and can result in a variety of complications, many of which

are associated with use of a heartlung machine. Inflammatory responses negatively affecting multiple organ systems have been observed in patients who were perfused
with a heartlung machine during traditional open heart
coronary bypass surgery (7). These responses are due to
reactions between circulating blood and material surfaces
present in the heartlung machine.
Minimally invasive direct coronary artery bypass is an
alternative to open heart surgery. A small 510 cm incision
made between the ribs replaces the sternotomy to gain
access to the heart. A retractor is used to separate the ribs
above the heart to maximize access to the operative site (8).
Heart positioners and stabilizers (Fig. 13) are used to
minimize the motion of the heart, allowing surgeons
to perform the procedure on a beating heart, eliminating
the need for the heartlung machine. Some stabilizers
grasp the heart with suction cups. Others use a fork like
device to apply pressure to the beating heart to keep it
steady for anastomosis of the graft (8). A thorascope and
small endoscopic instruments are used to visualize the
surgical site and perform the surgical procedure. The left
internal mammary artery is commonly used as the grafted
Figure 13. Heart positioner and stabilizer used in MIDCAB procedures. The positioner attaches to
the sternal retractor and holds the heart in position using suction. It provides greater access to the
blocked coronary artery. The tissue stabilizer, attached to the sternal retractor, straddles the
blocked artery and holds the suturing site steady. This allows surgery on a beating heart,
eliminating the need for a heartlung machine along with its associated complications.
(Courtesy of Medtronic, Inc., Minneapolis, MN.)
542
Stereotactic Radiosurgery
In this noninvasive procedure, focused external beams of
radiation are delivered to specific locations in the brain to
treat tumors (9). The accuracy of the delivery prevents
damage to healthy brain tissue. The patients head is
constrained in a mask or frame during the 3045 min
procedure.
Newer radiotherapy systems include robotic linear
accelerators for delivery of radiation at any angle to the
patients head, a beam shaper to match the shape of the
beam to the three-dimensional (3D) shape of the tumor,
and imaging equipment to provide real-time tracking of
tumor location and patient positioning (9).
Treatment of Benign Prostatic Hypertrophy
Figure 14. A PTCA catheter and stent. Stent and balloon in

collapsed configuration for insertion and placement into
coronary artery (a). Inflated balloon causing stent to expand (b).
Expanded stent after collapse of balloon and removal of catheter
(c). (Courtesy of Sorin Biomedica Cardio SpA, Italy.)
vessel to bypass the blockage of the coronary artery. The

MIDCAB results in fewer complications, less blood loss,
and shorter hospital stays.
Percutaneous Transluminal Coronary Angioplasty
with Stent Placement
The PTCA method is used to open a blocked, constricted
coronary artery instead of bypassing the blockage with a
graft. Under local anesthesia, a steerable balloon catheter
containing a stent mounted to the balloon (Fig. 14a) is
inserted into the patients femoral artery and guided to the
constricted coronary artery under fluoroscopy. Once in
position, the balloon is inflated to compress and flatten
the plaque against the arterial wall, creating a larger
opening for blood to flow through the coronary artery.
The balloon is constructed with materials of different
stiffness such that pressure from the inflating balloon is
radially applied to the constricted area, and not to tissue
outside of the constricted region (3). During balloon inflation, the stent is expanded to a larger diameter Fig. 14b,
which is maintained after deflation of the balloon. The
catheter is removed, and the expanded stent (Fig. 14c) is
left in place to maintain a larger opening and prevent
restenosis of the coronary artery.
Most coronary stents are made of stainless steel, nitinol,
cobalt chrome molybdenum alloy, or gold (3). Some stents
contain coatings to improve biological performance (biocompatibility and resistance to clot formation) and/or elute
drugs into surrounding tissues to prevent restenosis of the
coronary artery.
The PTCA method does not remove plaque from the
coronary artery; it flattens it so it no longer restricts blood
flow. Plaque removal methods involve lasers and rotational
cutting devices (3).
Benign prostatic hypertrophy causes the prostate to

enlarge. An enlarged prostate applies pressure to the prostatic urethra that can occlude the urethra, reduce urinary
flow rate, and make urination difficult. The enlarged prostate can be removed with an open surgical approach or less
invasive transurethral resection of the prostate (TURP).
The TURP procedure involves cutting and removing segments of the prostate through a cystoscope or with a
resectoscope inserted into the urethra, and can result in
complications, such as incontinence and impotence. Prostatic balloon dilators and prostatic stents inserted into the
urethra have been used to expand the narrowed urethra.
Transurethral laser incision of the prostate (TULIP)
has been used to cut and remove prostate tissue. In this
procedure, a catheter mounted laser facing radially outward toward the prostate delivers laser energy that cuts
through the enlarged prostatic tissue, relieving pressure
on the urethra.
Newer approaches to treating BPH include transurethral microwave therapy (TUMT) and transurethral needle ablation (TUNA) (4). These procedures use microwave
and radio frequency energy, respectively, to heat and
destroy unwanted tissue without actually removing the
tissue. The TUMT method involves insertion of a urethral
catheter containing a microwave antenna into the bladder
neck. Channels contained in the catheter carry cooling
water to prevent thermal damage to the urethra while
microwave energy is used to heat the prostate for 1 h.
The TUNA method uses a cystoscope to insert two shielded
electrode needles through the urethra and into the prostate. Radio frequency energy is delivered to the prostate,
heating the tissue and destroying it. Thermal energy
causes the prostate tissue to stiffen and shrink, relieving
the pressure on the urethra caused by the enlarged prostate. Both TUMT and TUNA deliver controlled thermal
energy to a targeted area to selectively destroy prostate
tissue (4).
NEW DEVELOPMENTS IN TECHNOLOGY FOR MINIMALLY

INVASIVE SURGERY
New devices and technologies are being developed and
clinically evaluated for use in MIS. Robots can be used
to enable surgeons to perform more intricate surgical
procedures than can be done with endoscopic devices (10).

In robot-assisted surgery, the surgeon operates a console to
control mechanical arms positioned over the patient. The
arms become an extension of the surgeons hands. While
observing the procedure on viewing screens (one for each
eye) that produce a 3D image, the surgeon uses hand,
finger, and foot controls to move the mechanical arms
containing interchangeable surgical instruments through
a small opening in the patients body (10).
Other new technologies for MIS include 3D video imaging, integrated robotic and surgical navigation systems,
devices for mechanical retraction of the abdominal wall
(eliminating the need for insufflation), and telerobotics (8).
OUTCOMES OF MINIMALLY INVASIVE SURGERY
Comparison of MIS to Conventional Approaches
Minimally invasive surgical procedures result in reduced
patient trauma, less postoperative pain and discomfort,
and decreased complication rates. Hospital stays and postoperative recovery times are reduced, resulting in lower
hospital costs and allowing patients to return to work
earlier.
Studies comparing various MIS procedures to their
traditional open surgical counterparts have been conducted. One such retrospective study compared the clinical and economic aspects of laparoscopic and
conventional cholecystectomies (11). Results of 153 consecutive traditional procedures and 222 consecutive
laparoscopic procedures performed in a German teaching
hospital were evaluated. Researchers found that although
laparoscopic cholecystectomies required longer operative
times (92 vs. 62 min), they resulted in fewer complications
(6 vs. 9), shorter hospital stays (3 vs. 8 days), and an 18%
decrease in hospital costs, when compared to traditional
procedures.
In a Canadian study, the direct costs of conventional
cholecystectomy, laparoscopic cholecystectomy, and biliary
lithotripsy were compared (12). Researchers concluded
that laparoscopic cholecystectomy provided a small economic advantage over the other two procedures and attributed this to a shorter hospital stay.
In the United States, hospital stays following laparoscopic cholecystectomies typically ranged from 3 to 5 days.
Now, many of these procedures are performed on an outpatient basis. This additional reduction in hospital stay
further reduces hospital costs, resulting in a greater economic advantage over the conventional procedure.
A study comparing results of 125 consecutive off-pump
coronary bypass (OPCAB) procedures to a matched, contemporaneous control group of 625 traditional coronary
artery bypass graft (CABG) procedures was conducted (7).
The OPCAB procedure is a beating heart procedure that
does not involve a heartlung machine. Partial sternotomies were used with some patients in the OPCAB group.
Researchers found that the OPCAB procedure resulted in a
lower mortality rate (0 vs. 1.4%), reduced hospital stays
(3.3 vs. 5.5 days), 24% lower hospital costs, and a reduced
transfusion rate (29.6 vs. 56.5%), when compared to the
traditional CABG procedure. Excellent graft patency rates
543
and clinical outcomes were also reported with the OPCAB

procedure.
In another study of 67 MIDCAB patients, it was found
that average hospital charges for MIDCAB were $14,676
compared to $22,817 for CABG and $15,000 for coronary
stenting (13). The significantly lower charges for MIDCAB
were attributed to shorter hospital stays, elimination of
perfusion expenses, and reduction in ICU, ventilation, and
rehabilitation times.
Limitations of Minimally Invasive Surgery
There are several problems and limitations associated with
MIS procedures. First, some minimally invasive surgical
procedures can take longer to perform than their more
invasive counterparts (10). Second, open surgical procedures allow the surgeon to palpate tissue and digitally
inspect for tumors, cysts, and other abnormalities. Tactile
feedback also assists in locating anatomical structures. The
inability of the surgeon to actually touch and feel tissue and
structures at the operative site limits the diagnostic ability
of some MIS procedures (10). Third, a two-dimensional
(2D) image is produced by the endoscope and video monitor. The resulting loss of depth perception combined with
restricted mobility of instruments through a small incision
make manipulation of endoscopic instruments challenging.
Fine hand movements are difficult due to the long distances between the surgeons hand and working ends of
these instruments. Fourth, there is a limit to the number of
instruments that can be used at one time through trocars
and working channels of a laparoscope. Finally, instrumentation for MIS procedures is more expensive.
Insufflation presents a small risk not associated with
open surgical procedures (1,10). If the gas pressure inside
the abdominal cavity becomes too high or there is a small
defect in a blood vessel, then a gas bubble can enter the
bloodstream and travel to the heart or brain, causing
unconsciousness or death.
Laparoscopic removal of cancer tissue carries a very
small risk of transporting cancer cells to other parts of the
body. As tumor tissue is removed through the laparoscope,
some cancer cells may remain in the body. They may be
displaced from their original location resulting in the
spread of cancer cells to other parts of the body (10).
SUMMARY
Minimally invasive surgery is made possible through the
use of specialized devices and technologies. These include
endoscopes, surgical instruments, imaging and navigation
systems, catheters, energy delivery systems, and other
devices. The MIS procedures tend to require more time,
are more difficult to perform than conventional procedures,
and present some unique risks not associated with conventional procedures. Compared to conventional open surgical
procedures, MIS procedures demonstrate similar clinical
efficacy and reduce trauma and disruption of body tissues
and structures. Patients experience less pain and discomfort, fewer complications, and shorter recovery periods.
Hospital stays are reduced, resulting in lower hospital
costs.
544
MOBILITY AIDS
BIBLIOGRAPHY
1. Ball K, Endoscopic Surgery. St. Louis: Mosby Year Book; 1997.
2. Holland P, The Fundamentals of Flexible Endoscopes, Biomedical Instrumentation and Technology. Association for the
Advancement of Medical Instrumentation, p 343348, Sep./
Oct. 2001.
3. Webster J G, ed. Minimally Invasive Medical Technology.
Institute of Phys Publishing Ltd., 2001.
4. Spera G, The Next Wave in Minimally Invasive Surgery,
Medical Device and Diagnostic Industry, Canon Communications, August 1998.
5. Gill B, Navigation Surgery Changing Medical Device Development, Medical Device and Diagnostic Industry, Canon
Communications, December 2004.
6. Scuderi R G, Tria A J, eds., MIS of the Hip and the Knee: A
Clinical Perspective; New York: Springer-Verlag, 2004.
7. Puskas J, et al. Clinical outcomes and angiographic patency
in 125 consecutive off-pump coronary bypass patients. Heart
Surg Forum May 1999;2(3):216221.
8. Spera G, The kindest cut of all, Medical Device and Diagnostic Industry, Canon Communications, July 1998.
9. Technology Trends: Radiation System Could Be an Alternative to Surgery, Biomedical Instrumentation and Technology,
Association for the Advancement of Medical Instrumentation, p. 18, January/February 2005.
10. Comarow A, Tiny holes, big surgery. U.S. News & World
Report, July 22, 2002.
11. Bosch F, Wehrman U, Saeger H, Kirch W. Laparoscopic or
open cholecystectomy: Clinical and economic considerations.
Eur J Surg 2002;168(5):270277.
12. Conseil devaluation des technologies de la sante du Quebec
(CETS). The Costs of Conventional Cholecystectomy, Laparoscopic Cholecystectomy, and Biliary Lithotripsy. Montreal:
CETS, 1993.
13. Oz M, Goldstein D, Minimally Invasive Cardiac Surgery.
Humana Press; 1999.
See also ENDOSCOPES;
GAMMA KNIFE; MICROSURGERY; STEREOTACTIC
SURGERY; TISSUE ABLATION.
MOBILITY AIDS
RORY COOPER
ERIK WOLF
DIANE COLLINS
ELIANA CHAVEZ
JON PEARLMAN
AMOL KARMARKAR
ROSEMARIE COOPER
University of Pittsburgh
Pittsburgh, Pennsylvania
ments, prolonging life spans, and the staggering growth

in the aging population. Because of this growth, the population of individuals who use mobility aids is sure to grow in
the coming decades.
The goal of issuing wheelchairs, prosthetics, walkers or
rollators to individuals with mobility impairments independence remains the top priority when prescribing one of
these devices, other main concerns include safety, not
causing secondary injury (i.e., pressure sores, carpal tunnel syndrome, rotator cuff tear), and the physical design of
the device (e.g., weight, size, ease of use). Research has
shown that manual wheelchair users commonly report
shoulder, elbow, wrist, and hand pain (2). Low back pain
and fatigue are also common secondary ailments experience due to exposure of whole-body vibrations (3). Safety
research shows that proper fitting and wheelchair skill can
reduce injurious tips and fall in wheelchairs (4).
In order to fully maintain an active lifestyle, including
recreational activities, participating in the community,
and going to work, transportation for people with mobility
impairments is essential. With the added technology that is
necessary to allow people to use public or private transportation, added safety features must also be included to
maintain the security of both the drivers and the passengers.
Besides performing normal activities of daily living,
sports, and recreational activity are an important physical
and psychosocial aspect of any human being. The case is no
different with people who use assitive technology. With
dramatic advances in technology, people with mobility
impairments are able to go places and participate in activities that were once nearly impossible.
In recent years, wheelchairs, prosthetics, walkers, and
rollators have been designed to be stronger, safer, lighter,
more adjustable, and smarter than in the past, through the
use of materials like titanium and carbon fiber, using
advanced electronics, and so on. The technology of wheelchairs, prosthetics, walkers, and rollators has improved
dramatically in the past 25 years due largely in part to the
increased demand of consumers, their loved ones and
others who assist consumers, and the people who recommend and issue these technology devices. The term"recommend is used because it is crucial that these devices,
especially wheelchairs and prosthetics, be issued by a team
of professionals to provide the highest levels of independence and safety, and that this team is centered around the
client. All of these components are important to the further
development of the technology and, in turn, they may
result in the increased independence of people with mobility impairments.
INTRODUCTION
The National Center for Health Statistics estimates that
12.4 million adults are unable to walk a quarter of a mile in
the United States, and that 28.3 million adults have moderate mobility difficulty (1). Approximately 4 million
Americans use wheelchairs, and about one-half of them
use their wheelchairs as their primary means of mobility.
About 1.25 million people wear a prosthetic limb due to
injuries, birth anomalies, and disease. Once fatal injuries
are now survivable due to advancing medical achieve-
CLINICAL ASSESSMENT OF MOBILITY

The ultimate goal and outcome for a clinical assessment of
mobility should drive toward a successful wheelchair,
prosthetics, walker, or rollator recommendation that
enhances the quality of life expectations and their effectiveness as reported by the consumer. Quality of life is
specific to and defined by each person and/or family receiving clinical services. The consumer, their family, and care
MOBILITY AIDS
givers, must be actively included in this process, as they

will be most affected by the choice of the mobility aid. Also,
people chose their mobility devices based on the features
available that facilitate activities or address needs (5), and
the clinician should be aware of the consumer preferences
and the features of various devices.
The complexity of some of the mobility device components combined at times with involved disease processes
can make it virtually impossible for a single clinician to act
independently when recommending assistive technology.
Therefore, involving an interdisciplinary team is recommended in the decision making process (6,7). This team,
with the consumer as an involved member and a variety of
rehabilitation professionals, includes a physiatrist or similarly trained physician who understands the importance of
Assistive Technology, addresses medical issues and assists
with mobility decisions; the Occupational or Physical
Therapist with RESNA (www.resna.org) certified Assistive
Technology Practitioner (ATP) credentials, who is the point
person for evaluation and prescription; and the Rehabilitation Engineering (with RE Training and RET Credential)
who is a technology expert with the ability to design
modifytune devices, and who also understands the capabilities and applications of various technologies. Another
important team partner is the Assistive Technology
Supplier (ATS) or Certified Rehabilitation Supplier (CRTS,
National Association of Rehabilitation Technology Suppliers,
www.narts.org), who provides or manages demonstration
equipment, does routine home and workplace assessments,
and orders, assembles, and delivers the equipment. All
team members involved in the mobility aid selection process should have knowledge about the technology available
on the market. Peer reviewed journal articles (Assistive
Technology, Archives of Physical Medicine and Rehabilitation, Journal of Rehabilitation Research and Development
etc.), magazine articles and commercial database sources
such as ABLEDATA (http://www.abledata.com/), or WheelchairNet (www.wheelchairnet.org) are good places to
research devices or to direct consumers who want to inform
and educate themselves.
Resources available to the team and its members
includes a defined and dedicated space for demonstration
equipment, assessments, and evaluations, access to common activities and tasks (e.g., ramps, curb cuts, bathroom,
countertop), an electronic tracking system to follow clients
and their technology, assessment resources (e.g., pressure
mapping, SMARTWheel, gait force plate, actigraph), IT
Resources (e.g., email, web, databases, medline, paging,
cell, wireless), and the facilitieshospital commitment to
continuing education.
A mechanism for Quality Measures for AT Clinics will
provide valuable feedback on performance quality and
areas in need of improvement. An important tool to measure patient satisfaction is the information gained through
a satisfaction survey provided to every patient, in order to
find out whether the goals and desired outcomes have been
met. Feedback on performance quality is provided through
tracking mechanism of primary clinician credentials (ATS,
ATP, RET), dedicated staffing levels, continuing education
(CEUs and CMEs), compliance with the commission on
Accreditation of Rehabilitation Facilities (CARF) AT Clinic
545
Accreditation, as well as tracking of continuous quality

improvement.
Assessment
The occupational or physical therapist conducts the initial
assessment process and obtains critical information about
the consumer and their environment. This part usually
involves a structured interview with the consumer and
then a physical motor assessment. Establishing a medical
diagnosis that requires the mobility aid is vital to assure no
ongoing medical problems exist that are not being adequately addressed. To properly specify a mobility device,
the intentions and abilities of the consumer must be ascertained (8,9). The intentions and abilities may include how
people perform tasks, where the deficits are, and how
mobility systems can compensate for the deficits to augment task performance. Some outcome tools that are clinically used to measure the functional impact of mobility aids
are the QUEST, the FEW and the Wheelchair Skills Test
(WST).
Additional necessary information includes type of insurance, method of transportation, and physical capabilities.
Also, if the consumer has been using a chair, historical
information about their current chair should be addressed,
including problems they are having. The mobility device
chosen should also be compatible with the public and/or
private transportation options available to the consumer,
such as a bus, car, or van. The regularity of the surface, its
firmness and stability are important, when, for example,
recommending a wheelchair in determining the tire size,
drive wheel location, and wheel diameter. The performance
of a wheelchair is often dictated by the need to negotiate
grades, as well as height transitions, such as thresholds
and curbs. The clearance widths in the environment will
determine the overall dimensions of the wheelchair. The
climates the chair will be operated in, and the need to be
able to operate in snow, rain, changing humidity and
temperature levels, and other weather conditions, are
important considerations as well.
A physicalmotor assessment of strength, range of
motion, coordination, balance, posture, tone, contractures,
endurance, sitting posture, cognition, perception, and
external orthoses is an important first step to obtain a
basic understanding of an individuals capacity to perform
different activities. The history likely provided significant
insight related to their physical abilities. To verify this, a
physical examination should focus on aspects of the consumer that (1) help justify the mobility aid, (2) help determine the most appropriate mobility aid, and (3) assure that
medical issues are appropriately addressed.
Once the examination documents the need, or potential
lack of need for the mobility device the remainder of the
examination can focus on the appropriate technology. This
is best assessed by giving the consumer the opportunity to
try equipment to determine how functional and safe they
maneuver/operate the device within the clinical space.
During this part of the assessment, the consumer and
family must be informed of the positive and negative
characteristics of devices and services. The team needs
to educate the consumer or family to participate in
546
MOBILITY AIDS
choosing the device that will meet their needs (Locus of

Control) and assure the provision of follow-up services.
The in-home evaluation conducted by the ATS verifies
that the device is compatible and will fit within the home
environment of the consumer; that may also included
recreational and work environment. Once the appropriateness of a device is established, final measurements are
taken. For many people, a few simple measurements can be
used to determine the proper dimensions for a wheelchair
(10). Body measurements are typically made with the
consumer in the seated position. A Rehabilitation Technology Supplier, therapist, or other member of the rehabilitation team often completes this.
MANUAL WHEELCHAIRS
When most individuals think of manual wheelchairs they
envision the boxy, steel framed, standard wheelchairs
commonly seen at airports and shopping malls. These
wheelchairs may be acceptable for transport of short distances, but are good for little else. Their heavy weight and
complete lack of adjustability makes them poor choices for
anyone using a manual wheelchair for an extended period
of time.
The development of the lightweight and ultralight
wheelchairs evolved in the late 1970s with a select few,
extremely motivated, manual wheelchair users choosing to
perform modifications on their own wheelchairs to make
them faster and easier to propel (11). After these initial
steps the demand became far greater for lightweight,
adjustable wheelchairs and several companies were created to meet that need.
Research conducted on the new lightweight and then
ultralight manual wheelchairs have quantitatively shown
the benefits over the heavier, nonadjustable standard style
wheelchairs. Cooper et al. (12) reported that when subjected to the fatigue tests described in the ANSI/RESNA
standards, ultralight manual wheelchairs were 2.3 times
more cost effective than lightweight wheelchairs and 3.4
times more cost effective than depot style wheelchairs.
The benefits of lightweight and ultralight manual
wheelchairs do not end with higher cost efficiency and
longevity. They are also crucial in preserving the upper
extremities of their users. Because of the constant use of
the upper extremities by manual wheelchair users, they
tend to experience secondary injuries such as joint pain,
repetitive strain injury, and nerve damage (2). Compared
to standard and lightweight wheelchairs, the ultralight
wheelchairs have two very distinct advantages when

attempting to prevent secondary injury in manual wheelchair users: the primary advantage is the lower weight
(Fig. 1).
Standard style wheelchairs tend to be greater than
36 lb(16 kg), while lightweight wheelchairs tend to be
3034 lb(14 18 kg) and ultralight wheelchairs 20
30 lb(914 kg). The second advantage presented by the
ultralight wheelchairs is adjustability. Lightweight wheelchairs do have some adjustability, but not to the extent of
an ultralight wheelchair. Masse et al. (13) showed that
moving the horizontal axle position toward the front of the
wheelchair and moving the seat downward created a more
efficient position for wheelchair propulsion and resulted in
less exertion without loss of speed. Although some studies
have been conducted to asses the importance of wheelchair
setup in reducing upper extremity injury in manual wheelchair users, the solution has not been clearly defined.
However, what is certain is that the adjustability and
weight of manual wheelchairs are crucial parameters when
selecting a wheelchair for a user that will be propelling for
extended periods of time.
Another recent addition to manual wheelchairs has
been suspension elements, such as springs or dampeners
to reduce the amounts of vibration transmitted to manual
wheelchair users. During a normal day of activity for a
wheelchair user, they encounter bumps, oscillations, and
other obstacles that may subject them to whole-body vibration levels that are considered harmful. VanSickle et al. (3)
demonstrated that when propelling over certain obstacles,
vibrations experienced at the seat of the wheelchair and
the head of the user exceed the safety levels prescribed by
the ISO 2631-1 Standard for evaluation of human exposure
to whole-body vibration (14). Wheelchair companies have
attempted to reduce the amounts of transmitted vibration
by adding suspension to manual wheelchairs. Research has
been done to evaluate effectiveness of suspension at reducing vibration. Results show that on average, suspension
does reduce the vibration levels, however, the designs are
not yet optimally effective and may not be as effective based
on the orientation of the suspension elements (15,16).
POWERED ASSIST WHEELCHAIRS
Often, people using manual wheelchairs are required to
transition to using powered wheelchairs or are at a level of
capacity where they must choose between the two. This
may be because of increased amounts of upper extremity
Figure 1. (a) Standard, (b) lightweight, and (c) ultralight wheelchairs.
MOBILITY AIDS
pain, or the progression of a disease such as Multiple

Sclerosis or Muscular Dystrophy. Recently, the development of Pushrim Activated Power Assist Wheelchairs
(PAPAWs) has provided an alternative for these users.
The PAPAW works through the use of a battery and a
motor mounted directly into the wheel. The motor provides
a supplement to the user so that very little force input from
the user will still afford normal momentum. Transitioning
from a manual wheelchair to a powered wheelchair may be
difficult both physically and psychologically for some people. Users may not want to modify their homes or their cars
to accommodate a powered wheelchair and may be used to
providing their own mobility. Although the PAPAW provides a good intermediate wheelchair for users who may
still benefit from a manual wheelchair, but do not have the
strength or stamina, it also has its disadvantages. The
added weight of the motor driven wheels dramatically
increases the overall weight of the wheelchair and can also
be difficult for users during transfers. Additionally, algorithms for the control of the PAPAWs are not yet refined
and can lead to difficulty propelling the wheelchair.
POWERED WHEELCHAIRS
Powered wheelchairs represent devices that can provide an
incredible amount of independence for individuals who
may have extremely limited physical function. Perhaps
even more than manual wheelchairs, having the proper
elements and setup of a power wheelchair is vital. The
lifestyle of the user, the activities in which they would like
to participate, the environments to which they will
be subjected, and ability level have all contribute to the
correct prescription of a powered wheelchair. Many adjustments can be made to any particular powered wheelchair to
specifically fit the needs of the user. Seating systems,
cushions, added assistive technology to name a few. This
section will focus on the characteristics that differentiate
certain powered wheelchairs from one another (Fig. 2).
Powered wheelchairs come in three drive wheel setups:
front, mid, and rear wheel. Each of these setups has
different advantages and shortcomings. Powered wheelchair users are each unique and have specific requirements
for their activities of daily living, such as maneuverability,
obstacle climbing, or driving long distances. Mid-wheel
547
drive wheelchairs tend to provide greater maneuverability

because the drive wheels are located directly beneath the
users center of mass. Front-wheel drive wheelchairs are
often associated with greater obstacle climbing ability.
Rear-wheel drive powered wheelchairs are good for speed
and outdoor travel, but may not provide the maneuverability or stability for some users. The different wheelchair
setups provide the means for users to achieve their goals.
For the user, the joystick is probably the most important
part of the powered wheelchair. However the standard
displacement joystick is not acceptable for all users. Current technologies have allowed almost any user to operate a
powered wheelchair as well as other assistive technologies,
such as computers. Even the smallest abilities of the user
can be translated to powered wheelchair mobility such as
foot control, head control, finger control, tongue control and
so on.
Like manual wheelchairs, powered wheelchairs also
have different classifications. Medicare defines powered
wheelchairs in three major categories: Standard weight
frame powered wheelchair (K0010), Standard weight
framed powered wheelchair with programmable control
parameters for speed adjustment, tremor dampening,
acceleration control and braking (K0011), other motorized
powered wheelchair base (K0014) (17). Pearlman et al. (18)
recently conducted a study examining the reliability and
safety of standard weight frame powered wheelchairs. Of
the 12 wheelchairs tested, only 3 passed the Impact and
Fatigue section of the ANSI/RESNA Standards (19,20).
Medicare has recently proposed to stop funding these
powered wheelchairs, recommending the addition of a
programmable controller that would place it in the second
category (K0011). However, this may not be acceptable
since the problems exist mainly with the drive train or
the frame of the wheelchair and these parameters would
not change. In order to adequately serve the interests of
powered wheelchair users, the frames, drive trains, and
control systems all need to perform within the standards
put forward by ANSI/RESNA.
WALKERS AND ROLLATORS
Some individuals may have the ability to ambulate,
however, they may get tired very easily or have visual
Figure 2. (a) Front-, (b) mid-, and (c) rear-wheel drive powered wheelchairs.
548
MOBILITY AIDS
Wheelchair Basketball
Figure 3. Robotic walker for users with visual impairments.
impairment problems that may result in a fall and injury.

Fuller (21) reported that 33% of community-dwelling
elderly people and 60% of nursing home residents fall each
year. Walkers and rollators represent useful assistive
technology devices for these individuals by lending support
and weight relief during mobility. They may have zero,
two, or four wheels and possibly have hand brakes. They
may also contain a small area for sitting if the user becomes
fatigued or a basket for carrying items for the user.
Some elderly persons, in addition to having mobility
impairment, also have a visual impairment. Recent research
has investigated a new robotic walker (Fig. 3) that through
the use of sonar and infrared (IR) sensors can detect obstacles
as well as provide guidance along a preprogrammed path (i.e.,
maneuvering around an assisted living home) (22).
Wheelchair users who play basketball may have various

diagnoses, such as paraplegia, cerebral palsy, amputations, post-polio syndrome, or a disabling injury. Participants are not required to use a wheelchair for their primary
means of mobility or in their activities of daily living. Prior
to the actual game, persons who want to play basketball
must have their player classification level determined by a
qualified referee. To equalize the capability of each team,
the classification levels of the competitors are matched
(26).
Whether players play zone or person-to-person basketball, the basic rules apply to both. Because different players
have varying degrees of disability, rules have been developed that all players need to abide by. Keep firmly seated in
the wheelchair at all times. A player may not use a functional leg or leg stump for physical advantage. An infraction of this rule constitutes a physical advantage foul (27).
Wheelchair basketball is similar to an everyday wheelchair, but incorporates features that enhance maneuverability (Fig. 4). Basketball wheelchairs are lightweight to
allow for speed, acceleration and quick braking. The wheelchair must have four wheels. Two large, rear wheels and
two front casters. The front casters are 2 in. (50 mm) in
diameter and typically made from hard plastics, similar to
the material used to make inline skate wheels. The rear
wheels must be larger than or equal to 26 in. (338 mm) in
diameter. The rear wheels must have handrims. Basketball wheelchairs use spoke guards made of high impact
plastic. These guards cover the rear wheel spokes to prevent wheel damage and illegal ramming and picking. The
spoke guards provide several benefits: First, spoke guards
can be used to pick up the ball from the floor. Using a hand,
the player pushes the ball against the spoke guard and rolls
it onto their lap, Second, spoke guards protect hands and
fingers from injury when reaching for the ball near another
players rear wheel. Third, they provide space to identify
team affiliations and sponsor names. Camber is an important feature of basketball wheelchair as well. Camber
is defined as"the angle of the main wheel to the vertical,
or as a situation in which"the spacing between the top
SPORTS AND RECREATION DEVICES

As quality of life receives more attention, sports and
recreational activities have become more important to
individuals with disabilities. Sport and recreational activity participation provides many benefits to individuals with
disabilities. Physical activity reduces or slows down the
development of cardiovascular disease as well as modifies
risk factors including high blood pressure, blood lipid
levels, insulin resistance, and obesity (23). In addition,
the development of muscular strength and joint flexibility
gained through regular exercise improves the ability to
perform activities of daily living (24). Regular exercise may
help reduce clinical depression and days spent as an
in-patient in a hospital, and may improve social interactions and prolong life expectancy (25). With the positive
benefits of sports, exercise, and recreational activities in
mind, the purpose of this section is to describe some of the
more popular sports played by individual with disabilities.
Figure 4. Players competing in a friendly game of wheelchair

basketball.
MOBILITY AIDS
points of the wheels may be less than the spacing between

the bottom points. Increasing camber slightly reduces the
height of the seat, while it proportionally increases the
wheelbase, which corresponds to the width of the wheelchair. In the same way, with negative camber, the center of
gravity of the occupied wheelchair moves backward. From
a practical point of view, increased wheel camber improves
hand protection as chairs pass through doors and, in terms
of basketball, camber makes a wheelchair more responsive
during turns and protects players hands when two wheelchairs collide from the sides, by limiting the collision to the
bottom of the wheels and leaving a space at the top to
protect the hands. Basketball wheelchair seats typically
have a backward seat angle slope of 5158. The angle of the
seat compared to the ground is known as seat angles.
Guards are an exception. Guards are allowed to have lower
seat heights and greater seat angles. These modifications
make chairs faster and more maneuverable for ball handling.
549
readily available (1) the upright and (2) the recumbent. In

an upright handcycle, the rider remains in an upright
position similar to the position the body takes when seated
in a touring bike. Upright handcycles use a pivot steer to
turn. Only the front wheel turns while the cycle remains in
an upright position. Transferring and balancing tend to be
easier on the upright cycle. In the recumbent handcycle,
the riders torso reclines and the legs are positioned out in
front of the cyclist. These cycles use a lean-to-steer
mechanism. The rider leans to turn, causing the cycle to
pivot at hinge points. Leaning to turn can be challenging if
the rider lacks trunk stability, in which case a pivot steering recumbent handcycle may be more appropriate.
Recumbent handcycles are lighter and faster, making them
the choice for hand cycle racing. Relatively minimal modifications are needed to accommodate individuals with
tetraplegia. Some of the modifications include hand cuffs
that can be mounted to the arm crank handles and elastic
abdominal binders which can be fitted around the user and
the handcycle seat to increase trunk stability.
Wheelchair Racing
Individuals with all levels of SCI as well as lower limb
amputees can participate in competitive races. The preferred racing chair among racers is the three-wheel chair
(Fig. 5). The three-wheel design is constructed from high
pressure tubular tires, light weight rims, precision hubs,
carbon disk/spokes wheels, compensator steering, small
push rings, ridged aluminum frame, and 2158 of wheel
camber. The camber in a racing chair makes the chair more
stable and allows the athlete to reach the bottom of the
pushrim without hitting the top of the wheels or pushrim.
Hand-Cycling
Cycling has been a popular outdoor sport for several years.
The adaptability of cycling to different terrains makes it a
favorite for many. Adaptive equipment for bicycles consists
of a hand cycle allows individuals with limited use of their
legs to utilize the strength of their arms (28). A handcycle
typically consists of a three-wheel setup to compromise for
the balance required when riding a two-wheeled bicycle.
Two-wheeled handcycles do exist but require a great deal of
skill and balance. Handcycle designs allow the user to
propel, steer, break, and change gears, all with the upper
extremities and trunk. Two types of handcycle designs are
Figure 5. Shows a racing three-wheeled wheelchair.
Wheelchair Rugby
Rugby is played indoors on a large gym on a basketball
court surface. Players use manual wheelchairs specifically
designed for the sport. Due to the level of contact, the chairs
have protective side bars on them and players are strapped
in to prevent injury. Most chairs are made of titanium or
steel to handle the hits that they sustain. In addition, the
low pointers have a high camber (angle of the wheels) so
that they can turn fast, as well as red push rim covers
so they can actually stick to the other persons chair and
hold them. The high pointers have armor on the front of
their chairs resembling a cow catcher so that they can push
through the other players without getting stuck (Fig. 6).
To be eligible to play rugby, players must have a combination of upper and lower extremity impairment. Most of
the players have sustained cervical level spinal injuries
and have some degree of tetraplegia. Like in basketball,
players receive a classification number based on there level
of impairment (29). Rugby consists of two teams comprised
of four players. The object of the game is for a player to have
possession of the ball and cross the opponents goal line.
Rugby wheelchairs are strictly regulated to ensure fairness. However, chairs may vary considerably depending on
Figure 6. Wheelchair rugby.
550
MOBILITY AIDS
a players preferences, functional level and team role.

Team roles may be assigned according to ability. Players
with upper body limitations tend to perform the defensive
blocking and picking roles. They use chairs that have
additional length and hardware. All rugby chairs
have extreme amounts of camber, 16208, significant bucketing, and antitip bars. The camber provides lateral stability, hand protection, and ease in turning. The bucketing
(knees held high relative to rear end) helps with trunk
balance and protection of the ball.
Tennis
Tennis players compete in both singles and doubles games.
Players are required to have a permanent mobility-related
physical disability that requires a wheelchair as the primary means of mobility. Tennis is played on the traditional
tennis court using the tradition size and height tennis net.
However, unlike traditional tennis, the ball is permitted
two bounces on the court before it must be returned. Brakes
are not permissible as stabilizers and the athlete must keep
one buttock in contact with the seat at all times.
Tennis players use a three-wheeled chair with a large
amount of camber to maximize mobility around the court.
The seat is situated at a steep backwards seat angle slope.
The angle helps with balance, keeps players against the
seat backs, and gives them greater control over the wheelchair. The knees tend to be flexed with the feet on the
footrest behind the players knees. With the body in a
relatively compact position, the combined inertia of rider
and wheelchair is reduced, making the chair more maneuverable (30). Handles and straps can also be added to the
chair. Many players incorporate plastic rigid handles into
the front of the seat. Players use these handles when
leaning for a shot or making quick directional changes.
Straps can be used around the waist, knees and ankles, to
help with balance (31).
Adaptive Skiing
Skis for skiers with disabilities have advanced state-of-theart skis that offer shock absorption systems, frames molded
to body shape, and quick release safety options. Skiers with
disabilities can maintain a similar pace to that of unimpaired athletes with the development of adaptive seating,
backrests, cushions, tethering ropes, roll bars and outriggers. Outriggers, an adapted version of a forearm crutch
with a shortened ski, provide extra balance and steering
maneuverability (32). Two types of sit-down adaptive skies
are available: Bi and Mono. Bi skis are appropriate for
skiers with limited trunk stability. With Bi skis, the skier
balances on two skies and angulates and shifts to put the
skis on edge. Bi skis have wider base of support, can
usually be mastered quickly with few falls and are easier
to control than a mono ski. The Mono Ski is the ski of choice
for individuals who want high end performance, maneuverability and speed. With a mono ski, the skier sits
relatively high on the seat of the ski over the snow. The
skier uses upper body, arm and head to guide their movement down the hill. Sit-, Mono- and Bi-skis have loading
mechanisms, usually hydraulic, that enable the individual
to raise themselves to a higher position for transferring

onto a ski lift.
TRANSPORTATION SAFETY AND ADAPTIVE DRIVING
FOR WHEELCHAIR USERS
Wheelchair users, like the entire population, use several
forms of transportation to travel from place to place: They
are passengers in public transportation systems (bus, subways, and vans) and private vehicles, and are potential
drivers of each of these types of vehicles. To ensure the
safety of all passengers and drivers, certain safety mechanisms must be in place: drivers must be able to safely control
the vehicle, and all seated passengers require seats
securely fastened to the vehicle and passenger restraints
(e.g., seatbelts) which can secure the passenger to the seat.
The requirement of passenger restraints is relaxed for
passengers in large vehicles, like busses and subways,
because of the low likelihood of high velocity crashes
(33). In many cases, adaptation of a vehicle is necessary
when the original equipment manufacturer (OEM) control
and/or securement mechanism cannot provide adequate
safety for wheelchair users because either (1) sensory and/
or motor impairments of the user requires adaptive driving
equipment so the vehicle can be safely controlled, or (2) the
user cannot safely or effectively use the OEM seat or
passenger restraint system in the vehicle.
Vehicle Control Systems
The complexity of the adaptive equipment required for safe
control of the vehicle is correlated to the type and level of
impairment of the wheelchair user. Adaptive driving
equipment can be as low tech as attaching a knob on a
steering wheel, and as high tech as fly-by-wire technology,
where computer-controlled actuators are added to all of the
controls, and the drive interfaces with the computer (via
voice and/or low or no-effort sensors). Adding actuators to
all of the driving and operating controls of the vehicle
through computer controls.
An example of this range is the types of steering adaptations available for people with disabilities. Figure 7 demonstrates both a typical steering knob for a person with little
to no loss of hand sensory-motor function (a), and (b) a knob
for someone with loss of some sensory motor function. Both
types of steering knobs serve the same purpose: They allow
the driver to safely steer the vehicle with one hand while
(typically) their other hand is operating a hand control
which actuates the fuelaccelerator and brake pedals. When
upper-extremity sensory-motor function does not allow for
safe turning of the OEM steering system (even with a
knob), actuators can be used in lieu of upper-extremity
function. These systems are named low- or no-effort
steering systems, depending on the type of assistance that
the actuators provide. Retrofitted controls for these systems are used and typically require removal of the OEM
equipment (e.g., the steering wheel and/or column). Consequently, when this level of technology is used, it usually
becomes unsafe for an unimpaired individual to drive the
vehicle (without significant training).
MOBILITY AIDS
551
Figure 7. Steering knobs.
Common fuelaccelerator and braking system hand controls are bolted to the OEM steering column, and actuate
each pedal with mechanical rods (Fig. 8). These types of
controls require nearly complete upper extremity function
to operate. When upper extremity function is substantially
impaired actuators are added to the braking and fuelaccelerator systems and are operated through some switching
methods. The types of switches depend on the most viable
control mechanism for the user: in some cases, a simple
hand-operated rheostat variable resistance switch (i.e.,
dimmer switch or rheostat) may be used, and in other
cases, a breath-activated pressure switch (sip-and-puff)
system may be used.
The above steering, fuelaccelerator, and brake adaptive
equipment are focused on the primary control systems of
the vehicle (those which are required to drive the automobile). Various adaptive equipment can control the secondary control system of the vehicle also (e.g., ignition
switch, climate control, windows). Like the primary control
adaptive equipment, equipment to modify the secondary
controls of the vehicle range from low to high tech. For
example, users with impaired hand function may require
additional hardware to be bolted to the ignition key so they
can insert the key and turn it to start the vehicle. Alternatively, hardware can be added to allow a drive to start
the vehicle via a switch, either remotely or from within the
cabin. Power windows and door-lock switches can be
Figure 8. Common hand controls.
rewired to larger switches, or in more accessible locations

for the driver.
Both primary and secondary control systems must be
placed in locations easily accessible to the driver. In some
cases, wheelchair riders will transfer out of their wheelchair directly into the OEM seating system, allowing for
most controls to remain in their OEM locations. If a wheelchair user remains in their wheelchair and drives the
vehicle the controls must be made accessible to their seated
position and posture. In these cases, along with the case a
wheelchair users remaining in their wheelchair as passenger, provisions must be made to safely secure the wheelchair to the vehicle, and to provide adequate passenger
restraints.
Wheelchair Tie-Down and Occupant Restraint
Systems (WTORS)
For both practical and safety reasons, when a wheelchair
user remains in wheelchair while riding in a vehicle they
must be safely secured to the vehicle. An unsecured wheelchair will move around, causing the user to be unstable
while the vehicle is moving. Being that power wheelchairs
can be in excess of 200 lb (91 kg) in weight, This instability
could cause harm to the wheelchair rider and/or the surrounding other passengers vehicle occupants. If the wheelchair user is driving, they may lose control of the vehicle if
they accidentally roll away from the vehicle controls.
Another important concern for an unsecured wheelchair
rider is in the case of an accident. An unsecured wheelchair
and user can easily be ejected out of the vehicle if they are
not secured. The WTORS systems are currently governed
by the ISO 7176-19 (34).
Several types of tie-down systems exist, including a
four-point belt system and various latching-type mechanisms which typically require hardware to be attached to the
wheelchair. The four-point belt systems are most common,
and are found on public busses, and also private vehicles.
Theses systems are the most widely used tie-down system
because they can be attached to a wide variety of wheelchairs. In some cases, manufacturers incorporate attachment rings for these tie-down systems into their
wheelchair. When no points of attachment are available
(most common situation), points at the front and rear of the
seat or frame be used. These attachment points must
be sufficiently strong to secure the wheelchair in the event
552
MOBILITY AIDS
of a crash, and be in locations which will allow the straps to

be oriented within a specified range of angles with the
horizontal (front straps: 30608, rear: 30458).
Unfortunately, tie-down systems are they are not convenient to use: a second person (other than the wheelchair
user) is typically needed to help secure the wheelchair,
making the operation laborious, and in some cases awkward for the wheelchair users who may not be comfortable
with another person touching their wheelchair or
encroaching on their personal space. Consequently, and
especially on public transportation, these systems are
commonly unused and the wheelchair user relies on their
brakes wheel-locks for stability, risking their own safety in
a crash.
Other mechanisms have been used to secure a wheelchair to the vehicle. These included wheel-clamps and t-bar
systems. With these mechanisms, wheelchairs are secured
to the vehicle through a mechanical clamp that adjusts to
the wheelchair size. These systems are quicker to attach to
the wheelchair, but are still difficult or impossible for a
user to use independently.
A variety of wheelchair tie-down systems have been
developed that allow the wheelchair users to independently lock and unlock their wheelchair to the vehicle. A
common one used for personal vehicles is the EZ-Lock
System, which is a hitch system for the wheelchair. This
system allows the wheelchair user to maneuver the wheelchair so a specialized hitch attached to the wheelchair is
captured into a latch bolted to the vehicle; both electric and
manual release mechanisms can be used to unhitch the
wheelchair from the device, allowing for custom placement
of a hitch for easy accessibility to the wheelchair user. A
drawback to this system is the specialized hardware that
must be attached to the wheelchair that restricts folding a
manual wheelchair and reduces ground clearance.
Because this system is designed to allow the user to
drive forward into the device, it works well and is common
in private vehicles where the wheelchair user drives the
vehicle. In larger vehicles, such as public busses, it is
typically more convenient for a user to back into a spot
and lock their wheelchair.
An ideal system for public transportation would be one
that a user can operate independently and that does not
require specific hardware to be attached to the wheelchair
that may not work on all wheelchair models. A system is
currently being developed at the University of Pittsburgh
that tries to achieve these goals.
Occupant restraint systems are the last requirement to
allow a wheelchair user to safely travel in a vehicle. These
restraint systems mimic the function of a seat belt, and can
be either attached to the wheelchair (integrated restraint)
or to the vehicle (Fig. 4). In both cases, the placement of the
restraints with respect to the body is critical to prevent
injury in a crasheither through direct insult of the
seatbelt with the body, or because of submarining (where
the torso slides down under the pelvic belt).
To ensure these WTORS systems and the wheelchair
themselves can safely survive a crash, standards testing is
in place. Rehabilitation Engineering and Assistive Technology Society of North America (RESNA), International
Standards Organization (ISO), and Society of Automotive
Engineers (SAE) have worked in parallel to establish

minimum standards and testing methods to evaluate
wheelchairs and WTORS systems (35). These tests mimic
those performed on OEM seat and occupant restraint
systems, which suggest the system should be able to withstand a 20 g crash (36). To encourage and guide wheelchair
manufacturers to build their wheelchairs to these standards researchers have developed a website to inform all
relevant stakeholders of the latest information (http://
www.rercwts.pitt.edu/WC19.html).
LOWER EXTREMITY PROSTHETICS

Prosthetics are devices that replace the function of a body
organ or extremity, unlike orthotic devices, which support
existing extremities. Prosthetics range from simple cosmetic replacements to complicated structures that contain
microprocessors for controlling hydraulic and pneumatic
components. Commonly used prosthetic devices primarily
include artificial limbs, joint implants, and intraocular
lenses. Approximately, 29.635.4% of the U.S. population
use prosthetic limbs (37) with >2% of them aged between
45 and 64 years using lower extremity (LE) prosthetics for
mobility (U.S. Bureau of Census, 2000). Amputation,
resulting from peripheral vascular diseases in the older
population (60 years and older) and trauma in young
population can be considered factors for the use of LE
prosthetics.
Research and development in clinical practice has
resulted in recent advances in the area of prosthetics
designs and controls technology. Examples of these
advances include the use of injection molding technology
for socket manufacturing, shock absorbing pylons, the
incorporation of neuro-fuzzy logic microprocessor-based
controllers for myoelectric prostheses, and microprocessorcontrolled prosthetic knees and ankles (38,39).
Prosthetic feet classified as"uniaxial allow for movement at a single axis in one plane, such as plantarflexion
and dorsiflexion of the ankle. In this type of prosthetic foot,
the heel is typically composed of the same density materials as the rest of the foot, with an option of different heel
height. Also, uniaxial feet have different options at the
rubber toe section in terms of flexibility, which depends on
the weight of the individual. Multiaxial prosthetic feet
(MPFs) have five degrees of motion in three planes: plantarflexiondorsiflexion, inversion/eversion, and rotation.
This feature provides stability to the user, while walking
on uneven surfaces and also aid in shock absorption lowering intensity of shear forces on residual limb. Elastomer
or rubberized material is used to alter, resist, or assist with
the different degrees of motion in the prosthetic foot. MPFs
also provide options for different heel heights [0.51 in.
(1326 mm)] and different degrees of toe material resistance, while split internal structures in the heel assist with
inversion/eversion on uneven ground. The MPFs are prescribed by the weight and shoe size of the consumer.
The solid ankle, cushion heel (SACH) prosthetic foot is
the most commonly prescribed prosthetic foot for lower
extremity amputations. The SACH foot is constructed out
of eurothene (plastic) materials with a less dense material
MOBILITY AIDS
553
Table 1. Functional Level and Devices
K0
K1
K2
K3
K4
Functional Level
Type of Device
No ability to ambulate or transfer safely; prosthesis

does not enhance mobility
Transfers and ambulates on level surfaces;
household use
Able to negotiate over low level environmental
barriers; limited community ambulation
Prosthetic usages are beyond simple ambulation;
able to traverse MOST environmental barriers and
is a community ambulator
Able to perform prosthetic ambulation exceeding
basic skills (i.e., high impact); child, active adult
and athlete
Cosmesis
incorporated at the heel. Use of materials with different

densities permits proper positioning while standing, as
softer heels aides in enhancement of the walking efficiency
after heel strike, by shifting center of gravity forward. A
device known as durometer is used to measure the density
of plastics used in prosthetic devices. The weight and
activity level of the individual using the prosthesis determines which heel density is selected, as heavier user
require firmer heel cushion. The stiffness of heel, also
determine amount of knee flexion and shock absorption.
Greater the heel stiffness more the knee flexion and lower
shock absorption during heel strike and vice versa. The
SACH foot also contains a keel made out of a hard wood or
composite material. Belting material is applied to the keel,
which prevent the keel it from breaking through the eurothene cover. During ambulation, the foot simulates plantar
flexion movement and prevents the loss of anterior support
during the push off at the toe.
Individuals who use foot prosthetic devices are assessed
for weight, potential activity levels, and type of use for
which they anticipate using their prosthetic devices. Based
on this assessment, clients are then categorized into four
functional levels:
Energy Storage and Return (ESAR) prosthetic feet are
fabricated to assist with the dynamic response of feet,
acting as a diving board from which a person can push
off during walking. These feet have capability to store
energy during stance phase and return it to the user to
assist in forward propulsion in late stance phase. The
ESAR has flexible keels and are prescribed by the anticipated activity level and weight of the person. Also, limited
evidence suggests use of ESAR as their use results in
increasing ambulation speed and stride length 713%
greater than with a conventional (SACH) foot in both
traumatic and vascular transtibial amputees (40).
Macfarlane et al. (40) compared energy expenditure of
individuals with transfemoral amputations who walked
with a SACH foot versus a Flex-Foot prosthetic. The SACH
has a solid ankle and cushioned heel construction, while
the Flex-Foot prosthetic has a hydraulic knee joint. The
authors determined that Flex-Foot walking resulted in
significantly lower exercise intensity, reduced energy
expenditure and improved gait efficiency. These findings
are significant considering the SACH foot is the most
commonly used foot prosthetic in the U.S. today (41). Lower
SACH
Low level energy storage feet
Energy storage prosthesis
Energy storage prosthesis
energy expenditure was also reported for individuals with

trans-tibial amputation with the use of Flex-Foot as compared with a SCAH foot.
Higher level of limb loss results in addition of more
prosthetic components. Prostheses for transfemoral amputations comprised of four basic components: the socket, the
knee joint, the pylon, and the foot. Pylons are classified as:
exoskeleton in which the weight of the individual is supported by the external structure of the prostheses (i.e., a
crustacean shank), or endoskeleton that is comprised of an
internal, weight-bearing pylon encased in moldable or soft
plastics (i.e., modular pylon). The knee mechanism use, a
conventional damping system, where a flow of (fluid or air)
is controlled by a valve and its operation is set for a
particular walking speed according to users preference.
The system described as intelligent prosthesis (IP), where a
diameter of damping controlling valve is changeable
according to varying speed of walking (42). Romo provided
guidance on the selection of prosthetic knee joints and
indicated that proper alignment impacts the effectiveness
of matched and adjusted knee joints for smooth and reliable
gait (43). Taylor et al. (44) compared effectiveness of an
intelligent prosthesis (IP), and pneumatic swing-phase,
control-dampening systems while walking on a treadmill
at three speeds of 1.25, 1.6, and 2 mph (2, 2.6, and
3.2 km h1). The results indicated lower VO2 consumption
for individuals using IP compared to controls-damping
system at 2 mph (3.2 km h1). The question often raised
by critiques concerns the cognitive demands by the high
end technology on the users. The results of the study by
Heller et al. (42) that investigated cognitive demand when
using the IP compared to a conventional prosthesis indicated no significant differences while using these prostheses for ambulation. Though not uncommonly
prescribed high rejection rates has been described for
prostheses after hip disarticulation and hemipelvectomy.
These prostheses consist of an addition of hip joint mechanism to other parts similar to prostheses prescribe after
transfemoral amputation.
Modular systems were first developed in the 1960s by
Otto Bock, which consisted of shock absorbing pylons that
contained with shock absorbers. Also, a reverse-pyramid
fixture at the ends of the pylon permits angular adjustments to the alignment of these devices with the residual
limb. Modular systems are lighter than the earlier wooden
554
MOBILITY AIDS
systems, allow for 158 of movement gain in either the

frontal or sagittal plane, and also permit internal and
external rotational adjustments. Modular systems can
extend the life of a prosthetic device, as worn parts can
be replaced. In addition, individuals using these systems
experienced less need for maintenance.
A significant improvement in the design procedure of
the prosthetics considers the interaction of forces between
prosthesis and residual limb can be found in the literature.
Jia et al. (45) studied the exchange of loads and forces
between the residual limb and prosthetic socket in transtibial amputation using the Finite Element Analysis (FEA)
method. Lee et al. (46) used FEA to determine contact
interface between the transtibial residual limb and prosthetic socket. The study determined the need for sameness
of shapes for both the residual limb and socket in order to
decrease peak normal and shear stresses over the patellar
tendon, anterolateral and anteromedial tibia, and popliteal
fossa. Winson et al. investigated the interaction between
socket and residual limb during walking using a FEA
model for transtibial prosthesis. Pylon deformities and
stress distribution over the shank were problems identified
during walking and results indicated need for pylon flexibility for better optimization and need of future studies
identifying fatigue life of these prostheses (47).
With advancement in the area of prosthetics designs and
development, simultaneous factors that need to be considered, use of these devices in clinical practice for targeted
population and cost containment. Premature abandonment
of mobility assistive devices, which might be due to poor
performance and/or changes in the need of the user, is not
uncommon and adds to the expense of these devices (48).
Improved quality of service delivery for LE prostheses, which
include identifications of reasons for successful use or nonuse of LE prostheses, is needed (49). Also, incorporation of
standardized performance testing procedure to ensure durability of LE prosthetics is vital to the appropriate prescription of, and satisfaction with, prosthetic devices.
Prosthetic devices of today incorporate advancements
from the aerospace and engineering fields and include the
use of new materials, such silicone elastomer gel sleeves in
to assist in the fit of prosthetic sockets, prosthetic feet made
from carbon-fiber composite components that are lighter in
weight, and surgical implantation of titanium prosthetic
attachment devices directly to bones of residual limbs
(50,51). Neuro- and microprocessors and sensors are now
incorporated on-board the prosthetic device to control knee
joint movement to improve the symmetry of different gait
patterns across a variety of cadence speeds. Hydraulic or
pneumatic devices are also used to dampen the swingthrough phase of walking with the prostheses to assist
with walking at difference cadences (52,53). Manufacturers are now using computer-aided design and manufacturing techniques to improve the fit of the prosthetic
sockets as well as component designs (54,55).
Because of the growing population of people in need of
mobility aids, and their demand to maintain their lifestyle,
whether that includes going to and from work, participating in extracurricular activities, or maneuvering around
their environment, continuing information must be gathered and disseminated to make these goals achievable.
Through technological advancements people who require

mobility aids can accomplish more of their goals than ever
before, however there are still people for whom the technology is not yet developed enough or cannot obtain the
proper devices to meet their needs. It is for this reason that
problems must continually be studied and innovations
must advance so that mobility aids will serve anyone
who requires them to meet their goals.
BIBLIOGRAPHY
1. Schiller JS, Bernadel L. Summary Health Statistics for the
U.S. Population: National Health Interview Survey, National
Center for Health Statistics. Vital Health Stat 10(220):2004.
2. Sie IH, Waters RL, Adkins RH, Gellman H. Upper extremity
pain in the postrehabilitation spinal cord injured client. Arch
Phys Med Rehab 1992;73:4448.
3. VanSickle DP, Cooper RA, Boninger ML, DiGiovine CP.
Analysis of vibrations induced during wheelchair propulsion.
J Rehab R&D 2001,38:409421.
4. Calder CJ, Kirby RL. Fatal wheelchair-related accidents in
the United States. Am J Phys Med Rehab 1990;69:184190.
5. Mills T, et al. Development and consumer validation of the
Functional Evaluation in a Wheelchair (FEW) instrument.
Disabil Rehab 2002;24(13):3846.
6. Chase J, Bailey DM. Evaluating potential for powered mobility. Am J Occup Therapy 1990;44(12):11251129.
7. Cooper RA, Cooper R. Electric Powered Wheelchairs on the
Move. Physical Therapy Products; July/August, 1998, p 2224.
8. Galvin JC, Scherer MJ. Evaluating, Selecting, and Using
Appropriate Assistive Technology. Gaitherburg (MD):Aspen
Publishers Inc.; 1996.
9. Cooper RA. Rehabilitation Engineering Applied to Mobility
and Manipulation. Bristol (UK): Institute of Physics; 1995.
10. Grieco A. Sitting posture: An old problem and a new one.
Ergonomics 1986;29:345362.
11. Cooper RA. A perspective on the ultralight wheelchair revolution. Tech Disab 1996;5:383392.
12. Cooper RA, et al. Performance of selected lightweight wheelchairs on ANSI/RESNA tests. Arch Phys Med Rehabil
1997;78:11381144.
13. Masse LC, Lamontagne M, ORiain. Biomechanical analysis
of wheelchair propulsion for various seating postitions. J
Rehab R&D 1992;29:1228.
14. International Standards Organization, Evaluation of Human
Exposure to Whole-Body VibrationPart 1: General Requirements. ISO 2631-1, Washington (DC): ANSI Press; 1997.
15. Wolf EJ, et al. Analysis of whole-body vibrations on manual
wheelchairs using a Hybrid III test dummy. Proceedings of
the annual RESNA conference; 2001. p 346348.
16. Kwarciak A, Cooper RA, Wolf EJ. Effectiveness of rear suspension in reducing shock exposure to manual wheelchair
users during curb descents. Proceedings of the annual
RESNA conference; 2002. p. 365367.
17. Centers for Medicare and Medicade Services (CMS), http://
www.cms.hhs.gov/providers/pufdownload/anhcpcdl.asp,
Accessed 2005. Jan 12.
18. Pearlman J, et al. Economical (K0010) Power Wheelchairs
Have Poor Reliability and Important Safety Problems: An
ANSI/RESNA Wheelchair Standards Comparison Study.
Proceedings of the Annual RESNA Conference; 2005.
19. American National Standard for Wheelchairs, Volume 2, Additional Requirements for Wheelchairs (Including Scooters) With
Electrical Systems. Virginia: Rehabilitation Engineering and
Assistive Technology Society of North America; 1998.
20. American National Standard for Wheelchairs, Volume 1,
Requirements and Test Methods for Wheelchairs (Including
Scooters). Virginia: Rehabilitation Engineering and Assistive
Technology Society of North America; 1998.
21. Fuller GF. Falls in the Elderly Am Fam Physician
2000;61(7):21592168.
22. Rentschler AJ, et al. Intelligent walkers for the elderly:
Performance and safety testing of the VA-PAMAID robotic
walker. J Rehab R&D 2003;40(5):423432.
23. Abel T, et al. Energy Expenditure in wheelchair racing and
hand biking- a basis for prevention of cardiovascular diseases
in those with disabilities. Eur J Cardio Prev Rehab
2003;10(5):371376.
24. Franklin BA, Bonsheim K, Gordon S. Resistance training in
cardiac rehabilitation. J Cardiopulm Rehab 1991;11:99107.
25. Kenedy DW, Smith RW. A comparison of past and future
leisure activity participation between spinal cord injured and
non-disabled persons. Paraplegia 1990;28:130136.
26. National wheelchair basketball association (20032004) official rules and case book retrieved from National Wheelchair
Basketball Association. http://www.mwba.org. Accessed.
27. Vanlandewijck Y, Daily C, Theisen DM. Field test evaluation
of aerobic, anaerobic, and wheelchair basketball skill performances. Int J Sports Med 1999;20(8):548554.
28. Janssen TWJ, Dallmeijer AJ, Van der Woude LHC. Physical
capacity and race performance of handcycle users. Jour
Rehab R&D 2001;38(1):3340.
29. Lapolla T. International rules for the sport of wheelchair
rugby. http://quadrugby.com/rules.htm. Accessed 2000.
30. Wheelchair tennis handbook. International Tennis Federation. http://www.itfwheelchairtennis.com. Accessed 2004.
31. Kurtz M. Difference makers. Sports N Spokes 2002;28(2):1014.
32. Russell JN, et al. Trends and Differential use of Assistive
Technology Devices: United States, 1994. Adv Data 1997;
292:19.
33. Shaw G, Gillispie B. Appropriate portection for wheelchair
riders on public transit buses. J Rehab R&D 2003;40(4):309
320.
34. International Standards Organization, Wheeled mobility
devices for use in motor vehicles. ISO 7176-19, Vol. 31. 2004.
35. Hobson D. Wheelchair transport safety - the evolving solutions. J Rehab R&D 2000;37(5).
36. Bertocci G, Manary M, Ha D. Wheelchair used as motor
vehicle seats: Seat loading in frontal impact sled testing.
Med Eng & Phys 2001;23:679685.
37. Mak AF, Zhang M, Boone DA. State-of-the-art research in
lower-limb prosthetic Biomechanics socket interface: a
review. J Rehab R&D 2001;38(2):161174.
38. Weir RF, Childress DS, Grahn EC. Development of Externally-Powered Prostheses for Persons with Partial Hand
Amputations. Proceedings of the Chicago 2000 World Congress on Medical Physics and Biomedical Engineering; 2000
July 23rd28, Chicago (IL):.
39. van der Linde H. A systematic literature review of the effect of
different prosthetic components on human functioning with a
lower-limb prosthesis. J Rehab R&D 2004;41(4):555570.
40. Macfarlane PA, et al. Transfemoral amputee physiological
requirements: comparisons between SACH foot walking and
flex-foot walking. J Prosthe & Ortho 1997;9(4):138143.
41. Nielsen DH, Shurr DG, Golden JC, Meier K. Comparison of
Energy Cost and Gait Efficiency During Ambulation in
Below-Knee Amputees Using Different Prosthetic Feet - a
Preliminary Report. J Prosthet Orthotics 1989;1(1):2431.
42. Heller BW, Datta D, Howitt J. A pilot study comparing the
cognitive demand of walking for transfemoral amputees
using the Intelligent Prosthesis with that using conventionally damped knees. Clin Rehab 2000;14(5):518522.
555
43. Romo HD. Prosthetic knee. Phys Med Rehab Clin N Am

2000;11(3):595607.
44. Taylor MB, Clark E, Offord EA, Baxter C. A comparison of
energy expenditure by a high level trans-femoral amputee
using the Intelligent Prosthesis and conventionally damped
prosthetic limbs. Prosthet Ortho Int 1996;8:116121.
45. Jia X, Zhang M, Lee WC. Load Transfer Mechanics Between
Trans-Tibial Prosthetic Socket and Residual Limb - Dynamic
Effects. J Biomech 2004;37(9):13711377.
46. Lee WC, Zhang M, Jia X, Cheung JT. Finite Element Modeling
of the Contact Interface Between Trans-Tibial Residual Limb
and Prosthetic Socket. Med Eng Phys 2004;26(8):655662.
47. Winson CCL, Zhang M, Boone D, Contoyannis B. FiniteElement Analysis to Determine Effect of Monolimb Flexibility on Structural Strength and Interaction Between Residual
Limb and Prosthetic. J Rehab R&D 2004;41(6a):775786.
48. Phillips B, Zhao H. Predictors of Assistive Technology Abandonment. Assis Technol 5(1):1993; 178184.
49. Scherer MJ. The change in emphasis from people to person:
introduction to the special issue on assistive technology.
Disabil Rehab 2002;24(13):14.
50. Marks LJ, Michael JW. Science, Medicine, and the Future:
Artificial Limbs. BMJ 2001;323(7315):732735.
51. Beil TL. Interface Pressures During Ambulation Using Suction and Vacuum-Assisted Prosthetic Sockets. J Rehab R&D
2002;39(6):693700.
52. Michael JW. Modern Prosthetic Knee Mechanisms. Clin
Orthop Relat Res 1999;361:3947.
53. Buckley JG, Spence WD, Solomonidis SE. Energy Cost of
Walking: Comparison of"Intelligent Prosthesis With Conventional Mechanism. Arch Phys Med Rehabil 1997;78(3):330333.
54. Twiste M, Rithalia SV, Kenney L. A Cam-Displacement
Transducer Device for Measuring Small Two-Degree of Freedom Inter-Component Motion in a Prosthesis. Med Eng Phys
2004;26(4):335340.
55. Lee WC, Zhang M, Jia X, Cheung JT. Finite Element Modeling of the Contact Interface Between Trans-Tibial Residual
Limb and Prosthetic Socket. Med Eng Phys 2004;26(8):655
662.
See also BLIND
AND VISUALLY IMPAIRED, ASSISTIVE TECHNOLOGY FOR;
ENVIRONMENTAL CONTROL; LOCOMOTION MEASUREMENT, HUMAN; REHABILITATION AND MUSCLE TESTING.
MODELING OF PHYSIOLOGICAL
SYSTEMS. See PHYSIOLOGICAL SYSTEMS MODELING.
MODELS, KINETIC. See TRACER KINETICS.
TARMO LIPPING
VILLE Ja NTTI
ARVI YLI-HANKALA
Tampere University of
Technology
Pori, Finland
INTRODUCTION
Anesthesia is one of the most complex and mysterious
phenomena in clinical work. The main feature of anesthesia
556
is the loss of consciousness, which suggests its relatedness

to sleep, epilepsy, and various kinds of brain trauma. In
the case of anesthesia and sedation, consciousness is
manipulated deliberately to prevent the patient from
being aware of their state and the medical procedures
carried through.
Recent decades have seen significant advancements in
mapping various psychological functions to corresponding
brain areas, however, the knowledge of the formation of
human consciousness is still based on uncertain hypothesis. This complexity makes anesthesia monitoring extremely challenging.
This article first addresses the problem of anesthesia
monitoring from the clinical, as well as from the physiological, point of view. The main emphasis is on monitoring
anesthetic depth as this is the most discussed topic in
anesthesia monitoring today. It starts with clinical indicators of anesthetic depth, gives an overview on the methods used in commercially available depth-of-anesthesia
monitors, and describes some new algorithms proposed
and evaluated for anesthesia electrocardiograms (EEG)
monitoring in recently published works. Finally, the feasibility of monitoring brain function is argued using neurophysiological parameters like EEG, Auditory Evoked
Potentials (AEPs), and so on, in the Intensive Care Unit
(ICU) and the Emergency Room (ER).
ANESTHESIA AS A PROCESS AND A PROCEDURE
Anesthesia can be seen from the clinical point of view as a
procedure, carried out according to certain protocol. From
the physiological point of view anesthesia is a process
evolving in the nervous system as the dose of an anesthetic
agent increases.
Anesthesia as a Procedure
The goal of general anesthesia in the operating room (OR)
is to render the patient unaware so that they do not feel
pain during the surgery or recall the events afterward. It is
also important that the patient does not react to surgical
stimuli by movement. In the ICU, the goal of sedation is to
keep the patient calm and painless. Too deep anesthesia
causes prolonged awakening times after surgery in OR and
longer treatment times in the ICU. The goals of anesthesia
and sedation can be achieved by hypnotics (unconsciousness producing drugs), analgesics (antinociceptive drugs),
and neuromuscular blocking agents. The choice of drugs is
mainly made based on experience and clinical signs during
the treatment.
Although general anesthesia is considered a safe procedure, various complications like postoperative nausea,
vomiting, and pain are relatively frequent. The incidence of
recall of events and awareness during anesthesia is rare
( 0.1%), however, the consequences may be traumatic for
the patient (1). Anesthesia-related mortality has decreased
significantly during past decades having recently been
estimated at 1 death per 200,000300,000 cases of anesthesia (24).
Anesthetic agents can be divided into inhalation anesthetics (e.g., halothane and isoflurane) and intravenous
anesthetics (e.g., thiopental and propofol). Intravenous

drugs are becoming more popular as they are short acting,
do not cause gas pollution, are easy to administer, and do not
cause airway irritation. Desirable properties of anesthetic
agents include rapid, smooth and safe induction of and
emergence from anesthesia, no accumulation in the body,
minimal effects on cardiovascular functions, no irritation to
tissues and veins, low potential to hypersensitivity reactions.
Anesthesia as a Process
During the last decades, it has become obvious that different
anesthetic and sedative agents produce their effects with
different mechanisms, and therefore from the physiological
point of view depth of anesthesia is a vague notion (5). It is
more meaningful to talk about components forming the state
we usually call anesthesia. These include amnesia, unconsciousness (hypnosis), antinociception, and neuromuscular
blockade (paralysis). Different neurophysiological modalities should be used in order to assess these components.
In the operating room, the patient is said to be anesthetized
while the term sedation is used in the ICU. Some drugs, like
propofol, are useful in producing both anesthesia and sedation at different concentrations, while some others are most
useful as anesthetics or sedatives. There is evidence that
anesthesia and sedation can be produced via different structures in the brainstem. Hypnosis and sedation cause similar
changes in the EEG signal (described in the section changes
in Neurophysiological Variables During Anesthesia). As all
the available depth-of-anesthesia monitors are either fully or
partly based on the EEG, the terms depth of hypnosis and
depth of sedation are used depending on the corresponding
clinical situation and are, in the context of available monitoring devices, roughly equivalent.
The action of anesthetics can be studied at various levels
of neuronal function (6). The models underlying these
studies can be divided into those operating at the molecular
cellular level and those explaining anesthetic processes at
higher levels (generator models). State-of-the-art knowledge on the molecular and neuronal substrates for general
anesthesia has recently been reviewed in Ref. 7. The model
proposed by Flohr describes the action of anesthetics as
disruption of computational processes dependent on the
NMDA receptors (8). The activation state of these receptors
in cortical neurons determines the complexity of representational structures that can be built-up in the brain and
thus the level of consciousness. Steyn-Ross et al. performed
numerical simulations of a single-macrocolumn model of
the cerebral cortex and found that the effects of anesthetics
can be modeled by cortical phase transition (9). Their
simulations explain several trends in the EEG caused by
anesthetic actions and predict the decrease in spectral
entropy of the EEG signal with deepening anesthesia. This
model has supported the development of the Entropy module, described in the section Recently Developed Methods,
Applied in Commercial Anesthesia Monitors. Great cautiousness must be taken, however, in interpretation of the
models operating at molecular level, because they include
only a small part of the neurophysiological functions
known to be involved in consciousness and generation of
anesthesia-induced EEG patterns.
557
REGULATION OF MEAN FREQUENCY

[6] Depress PFC
[5] Uncouple
parietal-PFC
interactions
Cortico-cortical / /
Cortical-subcortical
[2] Block interaction with DLPFC

to prevent memory transfer
CT-TC
loop
Dorsal striatum
[4] Block
gamma
loops
Thalamus
Cingulate
N. Reticularis
[3] Close
thalamic
gates
Hippocampus
Amygdala
S. Nigra
Sensory
inputs
Entorhinal CX
Ventral
tegmental area
Reticular
formation
[1] Depression of ARAS
GENERATION OF THETA
John et al. developed a higher level model based on a

complex neuroanatomical system described in (10). This
model, described and thoroughly discussed in volume 10
of Ref. 11, incorporates and explains an extensive bulk of
results obtained from EEG, evoked potential and magnetic
resonance imaging (MRI) image analysis, as well as laboratory studies. Loss of consciousness is described as the
following cascade of events, called the Anesthetic Cascade
by the authors, involving various brain regions (Fig. 1) (6):
(1) depression of the brainstem; (2) depression of mesolimbic-dorsolateral prefrontal cortex interactions leading to
blockade of memory storage; (3) inhibition of the nucleus
reticularis of the thalamus, resulting in closure of thalamic
gates (seen as increasing u rhythm in the EEG); (4) blocking of thalamocortical reverberations (g loop) and perception; (5) uncoupling of parietal-frontal transactions
(coherence in g frequency band decreases); (6) depression
of prefrontal cortex to reduce awareness (increase in frontal u and d rhythm).
Definitions of the EEG rhythms used in the description
of the Anesthetic Cascade are given in Table 1.
The model by John et al. underlies the Patient State
Index for depth-of-anesthesia monitoring, described in the
section Recently Developed Methods, Applied in Commercial Anesthesia Monitory. Although the Anesthetic Cascade model covers a large set of neurophysiological
functions, it does not explain patterns like burst suppression, for example.
Table 1. Definition of the EEG Rhythmsa
EEG Rhythm
Delta (d)
Theta (u)
Alpha (a)
Beta (b)
Gamma (g)
a
Septal nuclei
Frequency Range, Hz
<4
48
812
1225
2550
Exact frequencies may vary slightly from source to source.
Figure 1. The Anesthetic Cascade model

(redrawn with permission from: E. R. John
and L. S. Prichep, The anesthetic cascade:
A theory on how anesthesia suppresses
consciousness, Anesthesiology, Vol. 102,
Fig. 11 (p. 468), 2005).
MONITORING ADEQUACY OF ANESTHESIA

Clinical Indicators and Measures of Anesthetic Depth
The verb monitor originally means to check systematically
or to keep watch. Thus, monitoring actually does not
necessarily involve medical equipment, but refers also to
clinical inspection. As clinical indicators of anesthetic
depth are often used as a reference for automated depthof-anesthesia monitoring methods, they are shortly
described here.
In the case of inhalation anesthetics, drug concentration
can be monitored by measuring the partial pressure of the
anesthetic in exhaled air (end-tidal concentration). Due to
the variation of the potency of anesthetic agents, a universal unit of minimum alveolar concentration (MAC) has
been applied. The value 1 MAC is the partial pressure of an
inhaled anesthetic at which 50% of the unparalyzed subjects cease to express protective movement reaction to skin
incision. The primary rationale behind the development of
the term MAC was the need to compare the potency of
different volatile anesthetics, not the effort to monitor the
anesthetic state of an individual patient.
For intravenous anesthetics, no such direct measure can
be derived and the effect of anesthetics can be estimated
using pharmacokinetic models (effect-site concentration).
In this case, the accuracy of the estimate depends on the
adequacy of the model. If all subjects would react to anesthetics in exactly identical ways these concentration measures would provide a perfect indicator of adequacy of
anesthesia. However, there is an intersubject variability
in the effect of anesthetics, and therefore other indicators
are needed.
Clinical indicators of the adequacy of surgical anesthesia can be divided into those measuring hypnosis and
those measuring nociceptiveantinociceptive balance.
The indicators measuring hypnosis include pupillary light
reflex, tested by allocating a flashlight to one eye and
observing both pupils for constriction; corneal reflex,
558
Table 2. Ramsay Score for Assessment of Level of

Sedationa
Changes in Neurophysiological Variables

with Deepening Anesthesia
Score
Clinical Status
1
2
3
4
5
6
Patient anxious and/or agitated

Patient cooperative
Patient responds to commands only
Brisk response
Sluggish response
No response to loud auditory stimulus
All the commercial monitors of hypnosis employ the EEG

signal. Although different anesthetic agents induce specific
features and patterns in the EEG, certain common trends
in signal properties with deepening anesthesia can be seen.
At subanesthetic levels, several agents produce oscillations
at beta frequency range, sometimes called beta buzz. This
activity is seen dominantly in the frontal brain areas. With
increasing anesthetic concentrations, the activity becomes
more widespread, decreases in frequency and increases in
amplitude. Around concentrations, causing the subjects to
stop responding to stimuli (1 MAC for inhalation anesthetics), the EEG activity slows further and high amplitude
theta and delta waves occur. With further increasing concentration, the burst-suppression (BS) pattern occurs,
finally turning into continuous suppression. The dynamics
of this pattern, as well as the waveforms of bursts, varies
for different anesthetic agents (Fig. 2). Several anesthetic
agents tend to induce epileptiform seizure activity in
patients with a prior history of seizures and even in subjects with no previous history of seizures (15,16).
In addition to the EEG signal, AEPs have been used for
anesthesia monitoring. The latency of early cortical
responses Pa and Nb increases and the amplitude
decreases with deepening anesthesia (17). Also, late cortical responses to auditory stimuli, specifically the amplitude and latency of the N100 peak have been found to
improve the assessment of the level of consciousness in ICU
patients (18). A commercially available brain monitor, the
AEP Monitor/2 by Danmeter A/S, combines AEPs with
EEG parameters to calculate the cAAI index (see the next
section).
In most commercially available monitoring devices, the
EEG signal is obtained from the electrodes placed at the
forehead. This makes the recording procedure easier in
clinical situations. The electrodes tend to pick up frontal
EMG, which is an artifact from the point of view of the EEG
signal but may be used as a valuable indicator of nociception in light anesthesia (19). The EMG component of the
measurement is either explicitly or implicitly incorporated
into most of the available monitoring devices (see the
section Discussion).
Another neurophysiological variable proposed for
anesthesia monitoring is the respiratory sinus arrhythmia
(RSA) component of the heart rate (HR) signal (20).
Although potentially valuable addition to the assessment
of the level of consciousness, this variable has not made its
way to anesthesia monitoring devices to date.
See Ref. 13.
tested by applying a wisp of cotton wool to the cornea or

by electrical stimulation using special electrodes (12);
Eyelash reflex, tested by brushing the eyelashes with a
moving object or by electrical stimulation; loss of counting,
tested by letting the subject count slowly as long as they can
from the onset of infusioninjection; syringe dropping,
tested by letting the subject hold a syringe between their
thumb and forefinger as long as they can; loss of obeying
verbal commands.
The indicators measuring nociceptiveantinociceptive
balance include avoidance reaction to nociception. This is
mainly a spinal reflex, however, it correlates well with the
concentration of most anesthetics; electrical tetanic stimulation, applied using needle electrodes or adhesive skin
electrodes to the upper or lower limb; autonomic nervous
systemmediated reactions or motor reactions to laryngoscopy and endotracheal intubation. This is a natural stimulus in many clinical situations in the operating room.
These indicators test the functioning of different neural
pathways and their applicability depends on the anesthetic
used. For example, ketamine leaves corneal and pupillary
light reflexes intact.
For more graded and standardized clinical assessment
of sedation and hypnosis, several scoring systems have
been developed. Probably the most widely used such systems are the Ramsay score (Table 2) and the Observers
Assessment of Alertness and Sedation Scale (OAAS; Table 3).
These scoring systems are developed for use in the ICU as
they include scores for agitated states and cover mainly
lighter levels of anesthesia. Therefore, they do not necessarily indicate the adequacy of anesthesia for surgical
procedures. Also, the assessment obtained using these
scoring systems is subjective.
Table 3. OAAS Score for Assessment of Level of Sedationa
Score
Clinical Status
Responds readily to command

spoken in normal tone
Lethargic response to command
spoken in normal tone
Lethargic response to command
spoken loudly and repeatedly
Appropriate response to loud tone
and mildly painful stimulus
Appropriate response to loud tone
and moderately painful stimulus
No response
4
3
2
1
0
a
See Ref. 14.
Short History of Brain Monitoring in Anesthesia

Since the first measurements of human electroencephalogram, performed by Hans Berger in 1920s, this modality
has been applied to studying the effects of various drugs,
including anesthetics. The emergence of microprocessors
and digital techniques for signal analysis opened new
perspectives for anesthesia monitoring.
The first commercial brain monitoring device based on
digital signal analysis was the Cerebral Function Analyzing Monitor (CFAM1), developed in 1975 by Prior and
(a)
559
100 V
2s
Fz
(b)
100 V
7s
28
Figure 2. Samples of BS pattern in EEG during deep propofol (a) and sevoflurane (b) anesthesia.
Detection of BS suppression and calculation of BS ratio is an important part of all modern depth-ofhypnosis monitors. The pattern varies significantly among anesthetic drugs. In the case of propofol
anesthesia, spindles can be observed (marked by boxes in the figure). Note that the scale of the time
axes is different for upper and lower curves.
Maynard (21). This device used the Motorola 6808 8-bit

microprocessor. The display of the CFAM1 was divided into
two sections, one showing the 10th and 90th percentile as
well as the mean of the EEG amplitude distribution while
the other showing the percentage of weighted (prewhitened) EEG activity per herz in the beta, alpha, theta,
and delta frequency bands (Fig. 3). In addition, muscle
activity, EEG suppression ratio, and electrode impedance
were displayed. An important feature of the CFAM1 was
the possibility of monitoring averaged evoked potentials.
Since the introduction of CFAM1, the CFAM family of
brain monitors has been continuously developed with
the recently introduced CFAM4 being the latest member
of this product family. Comprehensive list of publications
referring to the CFAM family can be found at www.cfams.com/references/a4a.htm.
In 1982 Datex-Ohmeda (Helsinki, Finland) introduced
its first EEG monitor for anesthesia, the Anesthesia Brain
Monitor (ABM). Like in most of the later monitoring
devices, the location of the EEG electrodes in the ABM
monitor was on the forehead. The monitor displayed the
root-mean squared (rms) value of the EMG and the RMS,
as well as the zero-crossing frequency of the EEG signal.
The EMG and EEG signals were obtained from the same
electrodesbandpass filter of 65300 Hz was applied to
obtain the EMG while frequencies 1,525 Hz were used
to obtain the EEG. The ABM monitor is described in (22).
At the beginning of 1990s Thomsen et al. took a different
approach to anesthesia monitoring in their Advanced
Depth of Anesthesia Monitor (ADAM) (23). They divided
the signal into consecutive 2 s segments, applied a prewhitening filter, and derived 11 parameters: the rms value
and 10 correlation coefficients from each segment. Either
the values of the first 10 autocorrelation lags or the coefficients of the 10th-order autoregressive model were suggested as features. To create a set of reference classes, an
unsupervised repetitive hierarchical cluster analysis was
applied to the data bank of preannotated recordings of
halothane and isoflurane anesthesia. Six clusters were

defined, corresponding to anesthetic levels from drowsiness to very deep anesthesia. The classification was
adjusted according to the anesthetic agent used. Burstsuppression was detected separately and the suppression
ratio in 2 s segments was incorporated into classification.
Anesthetic depth was displayed as the class probability
histogram: A plot where each line represented the clusters
Figure 3. Layout of the screen of the CFAM1 monitor (with

permission from D. Maynard).
560
obtained for 10 s period of the recording. The clusters were

color coded. In spite of its advanced approach, ADAM was
never implemented in a commercial anesthesia monitoring
device.
Recently Developed Methods, Applied in Commercial
Anesthesia Monitors
The Bispectral Index Score (BIS), developed by Aspect
Medical Systems Inc. in 1997, marked a breakthrough in
anesthesia monitoring. The output of the BIS monitor is a
single number between 0 and 100 achieved by combining in
a nonlinear fashion from the following parameters (24):
relative beta ratio calculated in spectral domain as

log PP3047
, where P3047 and P1120 denote signal power
1120
in frequency ranges 3047 and 1120 Hz, respectively;
SynchFastSlow measure calculated in bispectral domain

0:547:0
as log PB40:047:0
, where B0:547:0 and B40:047:0 denote the sum
of magnitudes of the bispectrum values in the corresponding
frequency ranges; BS ratio. Bispectrum (the third-order
spectrum), is defined as the two-dimensional (2D) Fourier
transform (FT) of the third-order cumulant sequence c3(k1,
k2) of the signal:
Bv1 ; v2 FT
$ c3 k1 ; k2
If the direct current (dc) component of the signal has

been removed (as is usually the case), c3(k1, k2) equals to
the third order moment sequence m3(k1, k2), defined as:
m3 k1 ; k2 efsnsn k1 sn k2 g
where e{
} denotes expected value. Overview on the estimation of higher order spectra can be found in (25).
The weighting of the three parameters forming the BIS
depends on signal properties and is not disclosed. In light
anesthesia, relative beta ratio is dominating while SynchFastSlow measure becomes more important with deepening anesthesia. The function combining the parameters
was developed empirically, based on thousands of EEG
records. An important part of BIS is its careful artifact
rejection scheme, dealing with heartbeat artifacts, eyeblinks, wandering baseline, muscle artifacts, and so on
BIS has become very popular among anesthesiologists;
the bulk of literature dealing with the behavior of BIS in
various clinical situations, discussing its advantages as
well as disadvantages, incorporates more than 1000
papers. Comprehensive bibliography can be found on the
web-pages of Aspect Medical Systems Inc.
At the beginning of this decade Physiometrix Inc.
brought to market the PSA 4000 depth-of-hypnosis monitor, based on the Patient State Index (PSI) (26). The development of the PSI was based on a library of 20,000 cases of
EEG records. In addition, a library of surgical cases, a
library of artifacts and results from volunteer studies (for
calibration), were used. In PSI, the EEG signal is measured
from four electrodes: Fp1, Fpz, Cz, and Pz, with the reference at linked ear electrodes. Signal analysis is based on
power in standard EEG frequency bands (see Table 1) and
incorporates the calculation of the following parameters:
absolute power gradient between Fp1 and Cz leads in the
gamma frequency band (2550 Hz); absolute power

changes between Fpz and Cz leads in beta (1225 Hz)
and between Fpz and Pz; leads in alpha (812 Hz) frequency bands; total spectral power (0.550 Hz) at the
Fp1 lead; mean frequency of the total spectrum at Fpz
lead; absolute power in delta frequency band (0.54 Hz) at
Cz; relative power at Pz lead in slow delta frequency band;
The calculated parameters go through a mathematical
transformation that guarantees their Gaussian distribution in order to be rescaled into the Z-score (Fig. 4). The
Z-score sets the calculated parameters into relation with
the parameter values obtained for reference population
giving the percentage of the reference population that lies
more standard deviations away from the mean than the
calculated parameter (6). The Z-scored parameters are fed
into discriminant analysis with adaptive discriminant
functions. EEG suppression is detected separately: The
suppression ratio is included in the discriminant analysis.
The discriminant analysis yields the PSI index: a scalar
between 0 and 100 with higher level of hypnosis corresponding to lower PSI value.
The Narcotrend anesthesia monitoring system was
developed by a German group and first introduced in
2000 (27,28). This system has its roots in sleep analysis:
A five-stage sleep scoring system was further developed
into a system of 6 stages and 14 substages for levelof-hypnosis monitoring. These stages are mapped to a scale
of 0100 in the Narcotrend algorithm. The EEG signal is
derived from one or two electrodes; the most common
electrode location is on the forehead, however, according
to the authors other electrode locations are possible. The
signal is sampled at 128 Hz and prefiltered using lower and
upper cutoff frequencies of 0.5 and 45 Hz, respectively. The
principal idea underlying the method is similar to that of
the PSI: Several variables calculated from the EEG signal
are fed to discriminant analysis with separate detection of BS
(Fig. 5). The variables are classified as time- and frequencydomain ones and contain signal power, autoregressive coefficients, relative power in standard EEG frequency bands,
median frequency (the frequency dividing the signal spectrum into two parts of equal energy), spectral edge frequency
(SEF95, the frequency below which 95% of signal energy is
contained), and spectral entropy. The algorithm also contains plausibility check to ensure that the signal segment is
actually similar to a typical EEG sample of corresponding
stage and to detect patterns in the EEG signal untypical for
general anesthesia (e.g., epileptic activity). The detailed
algorithm of the Narcotrend index is proprietary.
Another EEG-based depth-of-anesthesia monitoring
device is the recently introduced M-Entropy module for
the Datex-Ohmeda S/5 anesthesia monitor. As the name
indicates, the method is based on the idea that the entropy
of the EEG signal decreases with deepening anesthesia.
Signal entropy can be defined and calculated in many
different ways (see also the next section) of which spectral
entropy is employed in the M-Entropy module. Spectral
entropy in the frequency range f1f2 is expressed as
S f1 ; f2
f2
X
fi f1
Pn fi log
1
P n fi
Fs = 2500
FP1
FPZ'
PZ
CZ
561
Fs = 250
Filter/
decimation
Mains
notch filter
Power spectrums
QEEG measures;
Power spectrum; sub-banding
Nonlinearity
4
EOG
evaluation
4
Magnitude
QEEG measures epoch buffer
Artifact
index
4
Stationarity
Compute mean of measures
4
Slew rate
Log transform measures
4
Cautery
Z-Score components
[mean, standard deviation]
4
Suppression
4
Compute wake/sedated
Discriminant functions based on
Z-Scored components
Compute discriminant PSI

based on ratio of
wake to wake + Sedated
Data editor
Suppression
ratio
2.5 s
Artifact free EEG
4
Frequency
transform
Arousal
observer
Discriminant
observer
Suppression
observer
Filter & decimate
PSI
where Pn(fi) is the normalized power spectrum of the
signal. S(f1, f2) is again normalized by log N(f1, f2), where
N(f1, f2) is the number of frequency components in the
range f1f2, to give a value between 0 and 1. In the original
version of the device, the analysis was performed on a
single EEG channel measured from the forehead. In this
derivation, muscle activity dominates over the EEG at
frequencies higher than 30 Hz. The algorithm of the
M-Entropy module, like that of the early ABM-monitor
by Datex-Ohmeda, employs these high frequency components to detect the early response of the patient to nociceptive stimuli. This is done by calculating spectral entropy
over two frequency ranges: 0.832 Hz (called state entropy)
and 0.847 Hz (called response entropy). The difference
between these two entropies indicates the contribution of
the EMG component to the response entropy. As in the other
described monitors, BS is detected separately. The details of
the algorithm (variable window length, obtaining the output
value in the case of BS, etc.) are described in (29).
The Danmeter AEP Monitor/2 (further development of
the A-Line monitor) employs the composite AAI Index,
Figure 4. Schematic of the calculation of the

PSI index. (Redrawn with permission from
D. R. Drover et al., Patient State Index:
Titration of delivery and recovery from
propofol, alfentanil, and nitrous oxide
anesthesia, Anesthesiology, Vol. 97, Fig. 3
(p. 88), 2002).
combining the middle latency auditory evoked potentials

in 2080 ms latency range, calculated from the 2565 Hz
bandpass filtered signal, with spontaneous EEG. The
purpose of combining the two modalities is to get a better
response to the lightening of hypnosis due to, for example
surgical stimuli (achieved by usind AEPs) while retaining
sensitivity during deep anesthesia (achieved by using the
EEG). The schematic of the cAAI index calculation is
presented in Fig. 6. Using evoked potentials poses a
problem in on-line monitoring due to the long delay
needed for obtaining the averaged response. This problem
has been solved in the cAAI calculation by applying the
autoregressive model with exogenous input (the ARX
model). The ARX model enables to calculate the response
to stimuli based on the average of 18 sweeps using the
average of 256 sweeps as a reference. The algorithm is
described in detail in (30) and compared with conventional evoked potential averaging techniques in Ref. 31.
In addition to AEPs, the cAAI index incorporates logarithmic EEG power ratio logP3047 =P1020 and the burst
suppression ratio. The EMG is extracted and monitored
562
Biomedical Instruments Inc). These monitoring devices come

in the form of a handheld wireless PDA-type tool, convenient
to use in a clinical situation. The CSM displays the Cerebral
State Index, calculated based on the 642 Hz bandpass filtered EEG, the EMG component calculated from the same
signal, but in 7585 Hz frequency range, as well as the burst
suppression ratio. The algorithm of the second version of the
SNAP index is described in (32). Two variables, the low
frequency variable LF (0.140 Hz) and the high frequency
variable HF (80420 Hz) are derived from a single frontal
EEG channel. The HF and LF are scaled to fit into the
intervals 0.01.0 and 0.0100, respectively. The SNAP index
is expressed as SI 100(HF LF); thus the index can be
thought of as the reversed version of HF-modulated LF.
EEG
Filters
Artefact
detection
Frequency-domain analysis:
Time-domain analysis
- fast fourier transform
- amplitude measures
- autoregressive parameters - spectral parameters
New Parameters Proposed for Monitoring Anesthetic Depth
Suppression detection /
discriminant functions
"Background"
parameters
Plausibility
checks
Narcotrend stage (A-F)

and
Narcotrend index (100-0)
Figure 5. Schematic of the calculation of the Narcotrend index

(with permission from B. Schultz).
separately based on the 6585 Hz bandpass filtered

signal.
A somewhat different concept of anesthesia monitoring
has been used in the Cerebral State Monitor (CSM; Danmeter
A/S, Odense, Denmark) and the SNAP monitor (Everest
A-Line Electrodes
BP filter
AEP
25-65 Hz
AMP
Estimate
SNR
In spite of the large selection of available methods, new

parameters for quantifying depth of hypnosis are being
proposed continuously. This is mostly due to the following
reasons: the variety of procedures and combinations of
drugs in surgical anesthesia is wide. No method performs
well in all cases; monitoring in anesthesia is closely related
to monitoring brain dysfunction and detection of brain
ischemia and hypoxia: important tasks faced in cerebral
function monitoring in the ICU and emergency room (see
the section Monitoring Outside the Operating Theater).
The available depth-of-hypnosis monitors are generally not
suitable for these applications; the neurophysiological
basis of consciousness is still an unsolved problem: applying modern signal analysis tools to neurophysiological
measurements during anesthesia can hopefully offer
new insight to the problem.
Several groups have recently published studies on the
behavior of various complexityentropy measures during
MTA256
sweeps
If SNR low
smooth signal
MTA18
sweeps
If SNR low
smooth signal
ARX
MODEL
Smooth
factor
AAI
calculation
Show SNR
bar/symbols
A/D
Converter
EEG
Signal
OK?
Yes
Bandpass filter
EMG
65-85 Hz
EMG
calculation
LF / HF
No
Reject
Bandpass filter
Burst Suppr.
1-35 Hz
BS %
calculation
Figure 6. Schematic of the calculation of cAAI index (with permission from E. W. Jensen).
anesthesia and sedation. These measures come from different signal analysis frameworks.
Correlation dimension is a measure for quantifying the
behavior of chaotic signals in the phase space (33). The
signal s of finite length N is divided into Nm1 time
series: sm(i) {s(i), s(i 1), . . . , s(i m1)}, where m is the
embedding dimension. After that, for each i the quantity
Cim(r) is calculated:
Cm
i r
number of such j that dSm i; Sm j r

4
Nm1
where the distance d between the phase space vectors sm(i)

and sm(j) is defined as
dSm i; Sm j
max jsi k 1 s j k 1j 5
563
Other measures of the EEG, found to correlate well with

depth of hypnosis, include LempelZiv complexity and
Higuchi fractal dimension (35,40). LempelZiv complexity
is calculated by transforming the signal into symbols and
calculating the reoccurrence rate of these symbols (41).
Higuchi fractal dimension is calculated as the average rate
of increase in the difference of signal amplitude values as
the separation between the samples increases in logarithmic scales (42).
These studies demonstrate that although different measures of entropy or complexity quantify different phenomena, many of them may correlate with concentrations of
selected anesthetics when electrode positions and signal
bandwidth are selected properly.
k1;2;...;m
Correlation dimension D can be estimated as

D
d logCm r
d logr
MONITORING OUTSIDE THE OPERATING THEATER

6
P
where Cm r i Cm
i r=N m 1: Although EEG cannot be considered strictly chaotic, except in the case of some
abnormal conditions, this measure has, for example, been
found to have good correlation with the end-site concentration of sevoflurane (34).
Probably the most intensively studied complexity/
entropy measure for the assessment of depth of hypnosis
is Approximate entropy (ApEn). In general, entropy measures information-richness, regularity and randomness of
a signal. The intuitive idea behind anesthesia monitoring
using signal entropy is that with deepening anesthesia
EEG becomes more regular and its entropy decreases.
Approximate entropy, like correlation dimension, is calculated in the phase space. First, fm(r) is defined based on
Cmi(r) in Eq. 4 as
Fm r
Nm1
X
1
logCm
i r
N m 1 i1
Approximate entropy is then defined as

A pEnm; r Fm r Fm1 r
Approximate entropy has been studied and compared

to other methods as an indicator of anesthetic depth in
(3537).
The classical entropy measure, introduced for information theory by Claude Shannon in 1948 (38), the
P Shannon
entropy (ShEn), is calculated as ShEn i pi log pi ,
where pi is the probability that signal amplitude obtains
the range of values ai. In practice, ShEn can be estimated
based on the histogram of the values of signal samples, and
therefore long signal segments are needed to achieve
smooth histograms. An important property of Shannon
entropy is that signal samples are considered as independent trials of some experiment, taking no notice on the time
order of the samples. Signals having equal probability for
all possible amplitude values have the highest Shannon
entropy. In Ref. 39, it has been found that Shannon entropy
of the EEG recorded between frontopolar electrodes
increases with increasing concentration of desflurane:
A behavior opposite to other entropy measures.
Development of digital EEG equipment, increase in processing speed and memory capacity, and advancements in
telecommunication technology have made cerebral function monitoring feasible in ICU and ER. Brain monitoring
in ICU and ER has much in common with monitoring in
anesthesia as the changes in the EEG caused by intoxication, metabolic disturbances and brain ischaemia are similar to those induced by general Anesthesia. Also, in the ICU
the assessment of depth of sedation is desirable. The
advantages offered by EEG monitoring in the ICU are
based on the following findings (43): EEG is tightly linked
to cerebral metabolism; EEG is sensitive to brain ischemia
and hypoxia; EEG detects neuronal dysfunction at a reversible stage; EEG detects neuronal recovery when clinical
examination cannot; continuous EEG provides dynamic
information; EEG provides useful information about cerebral topography.
However, from the monitoring point of view, the situation
in ICU and ER is a lot more complicated compared to that of
OR. The patients may need various medication having effect
on the EEG signal and misleading automated EEG analysis,
the clinical situation of the patients is often complex, and the
surrounding is hostile for interference-sensitive equipment.
In ICU, recordings often need to last for several days and
nights without disturbing the normal care of the patient. In
ER, the EEG recording equipment needs to be extremely
flexible and easy-to-use. In both situations the interpretation of the recordings poses a problem as no experienced
EEG readers are usually around. The solution to the last
problem is the usage of telecommunication protocols to
transfer the data for interpretation.
Although the above described depth-of-anesthesia monitoring methods are sometimes applied to sedation monitoring and even to the detection of brain dysfunction in
ICU, their performance in this situation is questionable. It
is difficult to differentiate between the effects of hypoxia,
ischemia and sedative drugs. The importance of having the
underlying raw EEG signal available for review to confirm
the significance of any trends and changes suggested by
automatic analysis methods, especially in complex situations like ICU, has been repeatedly emphasized (44,45).
A comprehensive brain monitor for ICU, especially for
neuroscience ICU, should also be able to detect epileptic
564
patterns in EEG and desirably have the option for synchronous video recording (46). All this suggests that an adequate
brain monitor for ICU or ER should be a much more complex
device than todays depth-of-anesthesia monitors.
DISCUSSION
Several considerations are appropriate concerning the
available commercial depth-of-anesthesia monitors. First,
different modalities should be used to assess the different
components of anesthesia (see the section Anesthesia as a
Process). Selecting EEG and AEPs as the basis for the
assessment, the primary component of anesthesia considered would be hypnosis.
But even in this case there still remain other physiologically separate end-points like subcortically controlled
reactions to nociceptive input (e.g., autonomic reactions),
increased muscle tone, and movement response to surgery.
Adding the fact that there are many anesthetic agents of
different cell-level actions and that the interplay of hypnotic and antinociceptive medication modulates the anesthetic state (47), we are left with a complex situation that
makes the comparison of the available algorithms for
anesthesia monitoring a real challenge.
Another difficulty in comparing the results obtained
with different methods is posed by the frequency band
used for the calculation. All the commercial methods operate at least partly in the frequency domain although BIS
applies third-order spectrum and in the Entropy module a
nonlinear transformation follows the calculation of the
power spectrum. For anesthesia, monitoring frequency
domain can roughly be divided into the following physiologically meaningful areas: d (and partly u) frequencies,
indicative of pre-BS deep anesthesia ( 0.56 Hz); a and b
frequencies; the EMG component, overlapping with the
EEG and extending to > 100 Hz.
The devices differ in the usage of d frequencies and in
the way the EMG component is incorporated. While most of
the methods employ frequency band starting from 0.1 Hz
(SNAP)0.8 Hz (Entropy), the Cerebral State Index and
cAAI by Danmeter do not make use of d rhythms. Several
devices like the A-2000 monitor by Aspect Medical Systems
Inc., the AEP Monitor/2 and the Cerebral State Monitor by
Danmeter as well as the PSA 4000 monitor by Physiometrix Inc. display the EMG power separately from their
corresponding depth-of-anesthesia indices. The frequency
band the EMG component is obtained from varies from
device to device, falling into the range from 65 to 110 Hz.
The SNAP index, the Entropy module and the Narcotrend
index incorporate the information on EMG activity into
their depth-of anesthesia indexes in different ways. In
SNAP, the high frequency band used is 80420 Hz, while the
other two monitors use frequencies up to 47 Hz. The various
entropycomplexity measures proposed for the assessment of anesthetic depth are sensitive to the prefilter
settings as well (40). Thus it can be concluded that while
comparing the performance of various algorithms, the
following matters should be considered: the properties of
the algorithm itself, the frequency band of the EEG signal
it employs, and the location of the EEG electrodes.
In the future, it seems to be inevitable that brain

monitoring becomes more common in ICU and emergency
room. There is a compromise between the simplicity of the
presentation of the output and versatility of the method.
Monitoring such complex phenomenon as anesthesia by a
single number is clearly an oversimplification. On the other
hand, a device requiring sophisticated configuration and
displaying a lot of parameters difficult to interpret gets
easily rejected by clinicians. Connecting the algorithms to
physiological models would certainly help the interpretation of the monitor output. Future will show if any of the
new approaches such as measures of signal complexity find
their way into the commercial devices. Operating in the
frequency domain has the advantage of long-term experience in EEG analysis by means of frequency domain methods. Another advantage is the solid signal processing
theory of frequency analysis. On the other hand, the theory
of nonlinear systems is developing rapidly having made
itsway to physiological signal analysis in various
applications.
BIBLIOGRAPHY
1. Myles PS, et al. Patient satisfaction after anesthesia and
surgery: Results of a prospective survey of 10811 patients. Br
J Anaesth 2002;84:610.
2. Committee on Quality of Health Care in America IoM. In:
Kohn L, Corrigan J, Donaldson M, editors. To Err Is Human:
Building a Safer Health System. Washington: National Academy Press 1999;241.
3. Chopra V, Bovill J, Spierdijk J. Accidents, near accidents and
6 36.
complications. Br J Anaesth 1978;50(10):1041U
4. Lagasse RS. Anesthesia safety: Model or myth? A review of
the published literature and analysis of current original data.
5. Kissin I. General anesthetic action: An obsolete notion?
Anesthol Analg 1993;76:215218.
6. John ER, Prichep LS. The anesthetic cascade: A theory on
how anesthesia suppresses consciousness. Anesthesiology
2005; 102:447471.
7. Rudolph U, Antkowiak B. Molecular and neuronal substrates
for general anaesthetics. Nature Rev Neurosci 2004;5:709
720.
8. Flohr H, Glade U, Motzko D. The role of the NMDA synapse
in general anesthesia. Toxicol Lett 1998;100-101:2329.
9. Steyn-Ross DA, Steyn-Ross ML, Wilcocks LC, Sleigh JW.
Toward a theory of the general-anesthetic-induced 24 phase
transition of the cerebral cortex. II. Numerical simulations,
spectral entropy, and correlation times. Phys Rev E 2001;
64:011918 (12 p).
10. Hughes JR, John ER. Conventional and quantitative electroencephalography in psychiatry. J Neuropsychiat Clin Neurosci 1999;11:190208.
11. Volume 10: Conciousness and Cognition 2001.
12. Mourisse J, et al. Electromyographic assessment of blink and
corneal reflexes during midazolam administration: Useful
methods for assessing depth of anesthesia? Acta Anaesthesiol
Scand 2003;47:593600.
13. Ramsay M, Savege T, Simpson B. Controlled sedation with
alphaxolene/alphadalone. Br J Med 1974;2:656659.
14. Chernik D, et al. Validity and reliability of the observers
assessment of alertness/sedation scale: Study with intravenous midazolam. J Clin Psychopharmacol 1990;10:244251.
15. Yli-Hankala A, et al. Epileptiform electroencephaogram during mask induction of anesthesia with sevoflurane. Anesthesiology 1999;91:15961603.
16. Vakkuri A. Effects of Sevoflurane Anesthesia on EEG Patterns and Hemodynamics. Ph.D. dessertation, University of
Helsinki, Finland, 2000.
17. Thornton C, et al. The auditory evoked response as an
indicator of awareness. Br J Anaesthology 1989;63:113115.
18. Yppa rila H. Depth of Sedation in Intensive Care Patients: A
Neurophysiological Study. Ph. D. Dessertation, University of
Kuopio, Finland, 2004.
19. Paloheimo M. Quantitative surface electromyography
(qEMG): Applications in anaesthesiology and critical care,
Ph. D. dissertation Acta Anaesthesiology; Scandinavian.
Copenhagen. Munksgaard; 1990; Vol. 34, (Suppl. 93),.
20. Wang DY, Pomfrett CJD, Healy TEJ. Respiratory sinus
arrhythmia: A new, objective sedation score. Br J Anaesthology 1993;71:354358.
21. Maynard D. Development of the CFM: The cerebral function
analyzing monitor (CFAM). Ann Anaesthology Francaise
1979;20:253255.
22. Edmonds HL, Paloheimo M. Computerized monitoring of the
EMG and EEG during anesthesia. An evaluation of the
anesthesia and brain activity monitor (ABM). Int J Clin
Monit Comp 1985;1:201210.
23. Thomsen CE, Rosenfalck A, Norregaard-Christensen K.
Assessment of anaesthetic depth by clustering analysis
and autoregressive modeling of electroencephalograms. Comput Methods Progr Biomed 1991;34:125138.
24. Rampil IJ. A primer for EEG signal processing in anesthesia.
25. Nikias CL, Petropulu AP. Higher Order Spectra Analysis.
Englewood Cliffs (NJ): PTR Prentice Hall; 1993.
26. Drover DR, et al. Patient State Index: Titration of delivery
and recovery from propofol, alfentanil, and nitrous oxide
anesthesia. Anesthesiology 2002;97:8289.
27. Schultz B, Schultz A, Grouven U. Sleeping stage based
systems (Narcotrend). In : Bruch HP. et al., editors. New
Aspects of High Technology in Medicine 2000. Bologna: Monduzzi Editore; 2000:285291.
28. Grouven U, Beger RA, Schultz B, Schultz A. Correlation of
Narcotrend Index, entropy measures, and spectral parameters with calculated propofol effect-site concentrations
during induction of propofol-remifentanil anesthesia. J Clin
Monit Comput 2004;18:231240.
29. Viertio -Oja H, et al. Description of the EntropyTM algorithm
as applied in the DatexOhmeda S/5Entropy Module. Acta
Anaesthesiol Scand 2004;48:154161.
30. Jensen EW, Lindholm P, Henneberg SW. Autoregressive
modeling with exogenous input of middle latency auditoryevoked potentials to measure rapid changes in depth of
anesthesia. Methods Inf Med 1996;35:256260.
31. Litvan H, et al. Comparison of conventional averaged and
rapid averaged, autoregressive-based extracted auditory
evoked potentials for monitoring the hypnotic level during
propofol induction. Anesthesiology 2002;97:351358.
32. Wong CA, Fragen RJ, Fitzgerald PC, McCarthy RJ. The association between propofol-induced loss of consciousness and the
SNAPTM index. Anesthology Analg 2005;100:141148.
33. Grassberger P, Procaccia I. Measuring the strangeness of
strange attractors. Phys D 1983;9:189208.
34. Widman G, et al. Quantifcation of depth of anesthesia by
nonlinear time series analysis of brain electrical activity.
Phys Rev E 2000;62:48984903.
35. Zhang XS, Roy RJ, Jensen EW. EEG Complexity as a measure of depth of anesthesia for patients. IEEE Trans Biomed
Eng Dec 2001;48(12):14241433.
565
36. Bruhn J, Ro pcke H, Hoeft A. Approximate entropy as an

electroencephalographic measure of anestetic drug effect during desflurane anesthesia. Anesthesiology 2000;92:715726.
37. Bouillon TW, et al. Pharmacodynamic interaction between
propofol and remifentanil regarding hypnosis, tolerance of
laryngoscopy, bispectral index, and electroencephalographic
approximate entropy. Anesthesiology 2004;100:13531372.
38. Shannon CE. A mathematical theory of communication. Bell
System Tech J 1948;27:379423.
39. Bruhn J, et al. Shannon entropy applied to the measurement
of the electroencephalographic effects of desflurane. Anesthesiology 2001;95:3035.
40. Anier A, Lipping T, Melto S, Hovilehto S. Higuchi fractal
dimension and spectral entropy as measures of depth of
sedation in intensive care unit. Proceedings of the 26-th IEEE
EMBS Annual International Conference (EMBC04), San
Francisco; Sept. 2004; pp. 526529.
41. Lempel A, Ziv J. On the complexity of finite sequences. IEEE
Trans Infor Theory 1976;IT22:7581.
42. Higuchi T. Approach to an irregular time series on the basis
of the fractal theory. Phys D 1998;31:277283.
43. Jordan KG. Continuous EEG monitoring in the neuroscience
intensive care unit and emergence department. J Clin Neurophysiol 1999;16:1439.
44. Scheuer ML, Wilson SB. Data analysis for continuous EEG
monitoring in the ICU: seeing the forest and the trees. J Clin
Neurophysiol 2004;21:353378.
45. Ja ntti V, Mustola S, Huotari AM, Koskinen M. The importance of looking at the EEG when presenting uivariate variables to describe it. Br J Anaesth 2002;88:739.
46. Hirsch LJ. Continuous EEG monitoring in the intensive care
unit: an overview. J Clin Neurophysiol 2004;21:332340.
47. Kern SE, X Gie, White JL, Egan TD. Opioid-hypnotic
synergy. Anesthesiology 2004;100:13731381.
48. Schultz B, et al. Der Narcotrend Monitor: Entwicklung und
Interpre-tationsalgorithmus. Anaesthesist 2003;52:11431148.
See also ANESTHESIA
MACHINES; BLOOD PRESSURE MEASUREMENT; ELEC-
TROENCEPHALOGRAPHY; OXYGEN ANALYZERS; SAFETY PROGRAM, HOSPITAL;

TEMPERATURE MONITORING.
MONITORING, AMBULATORY. See AMBULATORY

MONITORING.
MONITORING, FETAL. See FETAL MONITORING.

REED M. GARDNER
LDS Hospital and Utah
University
Salt Lake City, Utah
INTRODUCTION
The word monitor has a variety of meanings, depending
on the context. A monitor can be any device for checking on
or regulating the performance of a machine, aircraft, or a
patient. A patient monitor is usually thought of as something that watches, warns, or cautions if there is a lifethreatening event. A more rigorous definition of patient
monitoring is Repeated or continuous observations or
566
measurements of the patient, his or her physiological

status, and the functions of life support equipment for
the purpose of guiding management decisions, including
when to make therapeutic interventions and assessment
of those interventions (1). As a result, a monitor should
not only alert physicians and nurses to potentially lifethreatening events, but perhaps should also control
devices that maintain life. The primary emphasis of this
section deals with hemodynamic monitoring of the critically ill patient who is in an intensive care unit (ICU), but
many of the principles apply to all hospitalized patients.
Hemodynamic monitoring relates to monitoring of the
blood pressure and blood flow in the cardiovascular system.
The cardiovascular system consists of the heart, lungs, and
blood vessels, and has a most important function in maintaining life in complex animals, such as humans. Oxygen
and fuel must be transported from their source to the
individual cells that consume them. The resulting waste
products of metabolism must then be disposed of. Thus, the
heart and blood vessels transport nutrients to the body and
remove the waste products. Clearly, if this system does not
function properly, the organism could be compromised. As
a consequence clinically applicable methods have been
developed to assess the function of the cardiovascular
system. Hemodynamic monitoring is one part of this complex monitoring strategy. Typical parameters measured
when performing hemodynamic monitoring are heart rate
and rhythm, measured through analysis of the electrocardiogram (ECG), blood pressure measurements in various
locations in the cardiovascular system, and estimates of
blood flow usually using cardiac output as a measure.
THEORY
Hemodynamic monitoring permits minute-to-minute
surveillance of the cardiovascular system and provides
physiologic data to assist in diagnosis as well as to guide
therapy (25). The cardiovascular system consists of the
heart, lungs, and blood vessels that supply blood to the
body and return blood from the peripheral tissue.
It is beyond the scope of this section to describe the
detailed anatomy of the cardiovascular system. However,
to understand the principles of hemodynamic monitoring
knowledge of the functional aspects of the cardiovascular
system is essential.
HEART
The heart is made up of four chambers: the right atrium
and the right ventricle and the left atrium and the left
ventricle (see Fig. 1). The right atrium accepts blood from
the systemic circulation (head, arms, and legs) via the
superior and inferior vena cava. On atrial contraction
the tricuspid valve between the right atrium and right
ventricle opens and blood flows into the right ventricle. On
ventricular contraction the right ventricle pumps blood
through the pulmonic valve into the pulmonary artery
and to the lungs where oxygen is added and carbon dioxide
is removed. Blood flows from the lungs to the pulmonary
veins and then into the left atrium. On atrial contraction
SP
MP
DP
ARTERIAL
SP = 90 to 150 mmHg
DP = 60 to 80 mmHg
MP = 70 to 90 mmHg
RIGHT ATRIUM
(RA)
MP = 2 to 6 mmHg
RIGHT VENTRICLE
(RV)
SP = 20 to 30 mmHg
END DP = 2 to 6 mmHg
PULMONARY ARTERY
SP = 20 to 30 mmHg
DP = 8 to 12 mmHg
WEDGE = 6 to 12 mmHg
LEFT ATRIUM
(LA)
MP = 6 to 12 mmHg
LEFT VENTRICLE
(LV)
SP = 90 to 50 mmHg
END DP = 6 to 12 mmHg
Figure 1. Outline drawing of the heart with its chambers and

typical pressures (expressed in mmHg) for each area of the heart.
Note the pressures are systolic (SP), diastolic (DP), and mean
(MP), as shown on the arterial pressure waveform.
the blood flows into the left ventricle through the mitral
valve. On ventricular contraction the left ventricle pumps
blood through the aortic valve to the systemic circulation
(aorta and the systemic vasculature).
The basic contractile element of the heart is the sarcomere, which is composed of myofilaments, contractile proteins that interdigitate and slide along one another during
contraction. Shortening of the sarcomere is the functional
unit of heart contraction. Physiologic and pharmacologic
agents can change the contractile characteristics of the
sarcomeres. Rate and contractility of the heart are controlled by sympathetic and parasympathetic innervation,
as well as circulating catecholamines.
Control of Heart Performance
Mechanisms regulating cardiac (heart) output involve not
only factors controlling performance of the heart as a
pump, but also factors affecting the systemic vascular
system and its resistance. Typically, the heart can increase
its output to a level of almost five times its resting value.
There are two methods by which the heart regulates its
cardiac output in response to stress, injury, or disease: by
changing heart rate and stroke volume.
Heart Rate Control
Heart rate can be changed rapidly and is thus one of the
most effective ways for the heart to change its cardiac
output. For a healthy person, an increase in heart rate
can more than double the cardiac output when the heart
rate increases to near 180 beats
min1. However, if a
patient with heart disease increases their heart rate to
>120 beats
min1 they may have deleterious responses
because of the increased demand for oxygen by the heart
muscle. Blood flow in the heart muscle occurs primarily
during diastole (the relaxation phase of heart contraction).

Increasing heart rate decreases the time for cardiac circulation during diastole. In normal subjects, decreasing the
heart rate to 50 beats
min1 may not decrease cardiac
output because there is increased diastolic filling time that
increases stroke volume.
567
eject blood. Resistance (R) is calculated by measuring blood

flow and pressure and then using Ohms law {Eq. 1}.
R
mean blood pressure

cardiac output
(1)
Systemic Circulation
Stroke Volume Changes
The stroke volume of an intact ventricle is influenced by (1)
ventricular end-diastolic volume (called preload), (2) ventricular afterload, and (3) contractility.
Preload. Preload is the term used to define the enddiastolic stress in the wall of the ventricle. For example, zero
preload would result in the ventricle ejecting no blood. However, with increased preload, ventricular ejection generally
increases linearly until the capacity of the pump (heart) is
exceeded. Since the end-diastolic volume so profoundly
influences the myocardial fiber length it has a great influence
on the myocardial performance. The FrankStarling law
describes this principle and is illustrated graphically in
Fig. 2. The most accessible measure of right ventricular
preload is the right atrial pressure. Left atrial pressure is
used to estimate left ventricular preload. Since the left
ventricle does most of the work of the heart, it is usually
the first part of the heart muscle to fail. Consequently, the
measurement or estimation of the left atrial pressure is
important in assessing a patients hemodynamic status.
Afterload. Afterload is a measure of the impedance
(resistance) against which the right or left ventricles must
Blood flow to the periphery of the body is controlled by local

autoregulation and by the autonomic nervous system.
Local autoregulation of blood flow helps tissue meet its
oxygen requirements. For example, with decreased blood
flow, metabolic byproducts increase, causing local vasodilatation that tends to increase blood flow. There are baroreceptors, similar to blood pressure transducers, located
in the aortic arch and the carotid sinus which sense blood
pressure. Via the baroreceptor reflex mechanism, the body
regulates the blood pressure. In addition, chemoreceptors
in the carotid sinus and other locations regulate respiration by responding to changes in CO2 and O2.
Pulmonary Circulation
The pulmonary arterial vessels differ markedly from systemic arterial vessels; they have thinner walls, less muscle,
and have a resistance to blood flow about one-sixth that of
the systemic circulation.
Contractility. Contractility is a measure of how a
healthy heart performs. A healthy heart pumps vigorously
and nearly empties its ventricles with each beat and is said
to have excellent contractility. On the other hand, a compromised heart may not be able to empty effectively.
HEMODYNAMIC MONITORING
CARDIAC INDEX
HYPERDYNAMIC
NORMAL
Bedside hemodynamic monitoring makes use of data gathering procedures that were formerly only done in diagnostic cardiac catheterization laboratories. Understanding the
relationship between the pressure and blood flow in the
cardiovascular system is the primary reason for performing hemodynamic monitoring. The cardiovascular system
responds to many and varied stimuli and can be affected by
physical conditioning, drugs, disease, blood loss, injury,
and myocardial insult such as a heart attack. Because of
the complexity of factors controlling the body, it is necessary to make hemodynamic measurements on the system
to understand disease processes and provide optimum
therapy to the patient.
Electrocardiogram
FAILURE
PULMONARY WEDGE PRESSURE

Figure 2. FrankStarling curve of the heart showing the
ventricular performance (cardiac index) plotted against the enddiastolic volume typically estimated by using pulmonary artery
wedge pressure. Note to the right of these curves, there is a
pulmonary wedge pressure above which the heart in ineffective
in producing increased flow.
Electrocardiogram (ECG) monitoring is used to determine

heart rate and detect arrhythmias, pacemaker function,
and myocardial ischemia (lack of blood flow to the heart
muscle). To permit optimum ECG monitoring the signal
quality must be excellent (6). Since the ECG electrical
signal from the heart is only 0.52.0 mV at the skins
surface, it is best measured by properly preparing the skin
and optimally placing the electrodes. Skin can be properly
prepared by removing oil from the surface and abrading
the skin to remove the dry dead outer layer (stratum
568
granulosum). In 90% of patients, proper skin preparation

reduces electrode resistance from as high as 200 to as low
as 10 kV. Good electrode placement allows the electrodes to
receive the maximum ECG signal with minimum noise. By
placing the electrodes over bony prominence, such as the
sternum or clavicles, muscle artifact (EMG) can be
reduced. Motion artifact caused by movement of electrodes
can be minimized not only by proper skin preparation, but
also by taping a strain-relieving loop in the lead wires to
prevent movement artifact. Shielded wire on the ECG
leads helps minimizes pickup of alternating current (ac)
electrical fields from 60 Hz power sources, electrosurgical
units, and other sources like radio transmitters. The two
leads that connect the patient form a loop through which
magnetic fields pass and can induce unwanted voltages.
Pickup from magnetic fields can be minimized by decreasing the loop area, by keeping the lead wires close together
(usually twisted pairs), and by avoiding draping the lead
wires over motors, lights, or other electrically powered
instruments.
Electrocardiogram Arrhythmia Monitoring
Early in the development of monitoring techniques, the
application of computer technology to detect patterns of the
electrocardiogram caught the attention of those who
sought to improve care of the critically ill. The computer
appeared to be a logical candidate for relieving the nursing
and medical staff of the tedious chore of continuously
visually monitoring a multichannel oscilloscope.
Arrhythmia monitoring is one of the most sophisticated
of the bedside monitors tasks. People-based arrhythmia
monitoring is expensive and unreliable, and those who do it
find the task to be tedious and stressful. Today virtually
every bedside monitor has rhythm monitoring built in.
These monitors use computers and a variety of algorithms
to detect and classify ECG rhythm abnormalities. Classifying these rhythm abnormalities is important to hemodynamic monitoring since irregular rhythms can cause
dramatic inefficiencies in how the heart works as a pump.
For example, Fig. 3 shows three strip recordings of the
ECG and the corresponding pressure waveform from three
different arrhythmias (ventricular tachycardia, couplet,
which is two beats of abnormal electrical origin, and bigeminy where every other beat is from abnormal electrical
origin). Note that several of the abnormal beats are hardly
effective at creating any change in the arterial pressure.
Those same beats deliver small stroke volumes to the
patients systemic circulation. As a consequence, one cannot assume that the cardiac output remains constant or
increases just because the heart rate increases.
MEASUREMENTS
Blood Pressure Monitoring
Arterial blood pressure can be measured by both direct and
indirect methods. However, central venous pressure (CVP),
pulmonary artery (PA), and pulmonary capillary wedge
pressure (PCWP) used to estimate left atrial pressure, at
present, can only be measured by direct invasive methods.
Figure 3. Electrocardiogram and arterial pressure waveforms with

three different abnormal rhythms. (a) Ventricular tachycardia (VT),
which occurs during the last two beats of the strip. (b) Couplets where
two successive beats have an abnormal electrical origin. (c) Bigeminy
where every other beat is from an abnormal electrical origin.
Pressures are expressed in millimeters of mercury. For example,
the patient in (a) has a systolic arterial pressure of 109 mmHg,
a diastolic pressure of 53 mmHg, and a mean arterial pressure of
75 mmHg.
Arterial Blood Pressure: Indirect Measurement; Using a

Cuff. Recently, the American Heart Association has
updated its recommendations on accurate indirect
measures of blood pressure (7). The update reports that
the auscultatory technique with trained observer and
mercury manometer continues to be the method of choice
for measurement of a patients blood pressure in a physicians office. The report also suggests that hybrid devices
that use electronic transducers rather than mercury have
promise. The report indicates that oscillometric devices can
also be used, but only after careful validation.
Unfortunately, the indirect measurement of arterial pressure has serious limitations for patients in shock usually
signaled by low blood pressure. Also, since virtually all reliable
indirect pressure measurement techniques require cuff inflation, such measurements can only be made intermittently.
Direct Blood Pressure Measurements. The direct measurement of blood pressure allows for continuous and
accurate assessment of blood pressures. Direct and continuous pressure monitoring allows detection of dangerous
hemodynamic events and provides the information necessary to initiate and regulate patient therapy to prevent
catastrophic events. However, monitoring of pressures
provides valuable information only when it is obtained
in a technically satisfactory manner.
To accomplish direct blood pressure measurements, it is
necessary to insert a catheter directly into the cardiovascular
system (8). This invasive technique has risks that must be
weighted against the benefits that can be obtained. These
risks include, infection, blood loss, insertion site damage and
other factors (9,10). For many patients who are in shock or
who have cardiac disease, the benefits far outweigh the risks.
Formal methods for assessing these risks have been published by the Coalition for Critical Care Excellence (11).
Blood pressure can be measured on both the pulmonary
(right heart) and systemic (left heart) sides of the circulatory
569
system. Measurements of both pulmonary and systemic

parameters yield different and important cardiovascular
status. The CVP reflects the patients circulating blood
volume or venous tone, and right atrial and ventricular
pressures (right ventricular preload). To measure the right
atrial pressure accurately a catheter must be placed in a
major vein within the chest or directly in the right atrium.
The CVP values fluctuate about atmospheric pressure. The
level of the right heart is usually taken as the zero reference
point from which all other blood pressures are measured.
The CVP gives an indication of only the function of the right
heart, and not left hearts performance.
To measure the left atrial pressure, it is necessary to place
a catheter tip through the atrial septum from the right
atrium (usually done only with fluoroscopic control in the
cardiac catheterization laboratory) or estimating it by placing a pulmonary artery (SwanGanz) catheter in the
wedged position by inflating its balloon near the catheter tip.
EQUIPMENT
Components of Direct Pressure Monitoring Systems
The components of a direct blood pressure monitoring
system for critically ill patients are shown in Fig. 4 (6,8).
The components numbered 17 in the figure are known as
Figure 4. The 10 components used to monitor direct blood pressure. The monitoring components
are nearly independent of whether the catheter is in an artery (radial, brachial, or femoral) or in the
pulmonary artery. Size of transducer and plumbing components are enlarged for illustration
purposes. [Reproduced from Ref. 6, with permission.]
570
the plumbing system and must always be sterile because

the fluid contained therein comes in direct contact with the
patients blood. Today virtually all of these components are
disposable or single-use items to minimize patient infection. Components 811 in Fig. 4 are used for processing and
displaying pressure waveforms and derived hemodynamic
parameters.
1. Catheter. Arterial and pulmonary artery catheters
provide access to the patients blood vessels to
(a) monitor intravascular pressure and (b) provide
a site for samples for blood to allow determination of
blood gas and other laboratory testing parameters.
These catheters are typically placed by the
percutaneous method, either by the Seldinger
over-the-needle technique or by introducing the
catheter through a needle (8).
2. Sampling stopcock. Stopcock 1 is used as a site for
withdrawing blood for analysis. When filling the
catheter-tubing-transducer system with fluid,
precautions must be taken to be sure all central
switching cavities of the stopcock are filled and that
entrapped air bubbles are removed. Because stopcocks are especially vulnerable sources of patient
contamination, these devices must be handled with
extreme care; ports not in active use should be covered with sterile caps and medical personnel should
never touch open ports of the stopcocks.
3. Pressure tubing. The catheter and stopcock are normally attached to a continuous flush device and transducer by noncompliant pressure tubing. To optimize the
dynamic response of the catheter-tubing-transducer
system, long lengths of tubing must be avoided.
4. Stopcock 2. This stopcock is usually put in place to
allow disconnection of the flush device and transducer from the patient when the patient is moved or
when initially filling the system with fluid.
5. Continuous flush device. This device is used not only
when initially filling the pressure monitoring system,
but also to help prevent blood from clotting in the
catheter. These devices provide a continuously
flush of fluid at a rate of from 1 to 3 mL
h1.
6,7. Transducer dome and Pressure transducer. Today
virtually all transducers used for monitoring are
highly reliable, standardized, disposable devices
(12,13).
8. Amplifier system. The output voltage required to
drive an oscilloscope or strip-chart recorder is provided by an amplifier system inserted between the
transducer and display. Pressure transducer excitation is provided either from a direct current (dc) or ac
source at a voltage of 48 V revolutions per second
(rms). Most amplifier systems include low pass filters
that filter out unwanted high frequency signals. Pressure amplifier frequency response should be flat from 0
to 50 Hz to avoid pressure waveform distortion.
9. Oscilloscope. Pressure waveforms are best visualized
on a calibrated oscilloscope or other similar display
panel.
10. Digital processing and display. Digital displays

provide a simple method for presenting quantitative
data from the pressure waveform. They are found
on most modern pressure monitoring equipment.
Systolic, diastolic, and mean pressure are typically
derived from the pressure waveforms.
11. Strip-chart recorders. Frequently, strip-chart recorders are used to document dynamic response characteristics, respiratory variations in pulmonary
artery pressures, and aberrant rhythms and pressure waveforms.
STATIC CALIBRATION
Zeroing and calibrating the transducer are two important
steps in setting up the direct pressure-monitoring system.
Zeroing the Transducer
The accuracy of blood pressure readings depends on establishing an accurate reference point from which all subsequent measurements are made. The patients midaxillary
line (right heart level) is the reference point most commonly used (14). The zeroing process is used to compensate
for offset caused by hydrostatic pressure differences, offset
in the pressure transducer, amplifier, oscilloscope, recorder, and digital displays. Zeroing is accomplished by opening an appropriate stopcock to the atmosphere and aligning
the resulting fluid-air interface with the midaxillary reference point.
Once the system is zeroed the stopcock can be switched
to allow the patients waveform to be displayed. Pulmonary
artery and pulmonary artery wedge pressure are especially
susceptible to improper zeroing and should be measured
only after the zero point has been verified.
Sensitivity Calibration
The sensitivity of most pressure transducers is fixed at
5.0 mV
V1 of excitation applied per 1 mmHg (0.13 kpa)
and calibrated by the manufacturers to within 1%. This
degree of accuracy is adequate for clinical purposes. As a
consequence standardized transducers need only to be
zeroed to obtain accurate pressure measurements (12,13).
CHECKING DYNAMIC RESPONSE
In the critical care setting, where most hemodynamic
monitoring is carried out, the catheter-tubing-transducer
systems used can usually be characterized as an underdamped second-order dynamic system analogous, for
example, to a bouncing tennis ball. A second-order dynamic
system can be expressed mathematically by a second-order
differential equation with characteristics determined by
three mechanical parameters: elasticity, mass, and friction. These same parameters apply to a catheter-tubingtransducer system where the natural frequency (fn in Hz)
and damping coefficient determine the dynamic characteristics for a catheter-tubing-transducer system. For an
underdamped second-order system fn and define the systems
DAMPING COEFFICIENT
450
AMPLITUDE RATIO %
400
350
300
= 0.2
250
200
= 0.5
150
100
50
= 2.0
2
= 0.7
3
10
12
HARMONIC
(HR = 120)
0
0
14
16
18
Figure 5. Family of frequency versus amplitude ratio plots for five

different damping coefficients z and natural frequencies fn of
the plot shown is 10 Hz. When z 0.1, the system is very
underdamped, and when z 2.0, it is overdamped. The dashed
line shows the frequency versus amplitude characteristic
that would occur if the system had a flat frequency response.
Along the frequency axis are plotted the harmonics of the
pressure wave if the heart rate were 120 beats
min1 (2
beats
s1). Note that by the fifth harmonic (10 Hz) if z 0.1 the
true signal would be amplified five times. If z 2.0 there would be
an attenuation to about one-fourth of the amplitude. In both cases
there would be a gross waveform distortion because neither
situation reflects a high fidelity system dynamic response.
Fidelity of the system can be improved by increasing the fn or
adjusting z to be in the range of 0.50.7. [Reproduced from Ref. 6,
with permission.]
dynamic characteristics. In the clinical setting fn and can be

measured easily and conveniently by using the fast-flush
method.
Dynamic response characteristics of catheter-tubingtransducer systems have been defined by two interrelated
techniques. The first technique specifies the frequency
bandwidth and requires that the system frequency
response must be flat up to a given frequency so that a
specified number of harmonics usually 10 of the original
pulse wave can be reproduced without distortion (Fig. 5).
The second technique specifies fn and The plot of fn and
in Fig. 6 has five areas (6,15). If the characteristics of the
catheter-tubing-transducer plumbing system fall in the
adequate or optimal area of the graph, the pressure waveforms will be adequately reproduced. If the characteristics
fall into one of the remaining three areas, there will be
pressure waveform distortion. Most catheter-tubing-transducer systems assembled under optimal conditions are
underdamped, but a few fall into the unacceptable areas
of the chart. Methods for optimizing the catheter-tubingtransducer system components have been outlined (1520).
In the clinical setting, there are dramatic differences
between each patient setup; therefore it is mandatory to
verify the adequacy of each pressure-monitoring system by
testing them. The testing can be done easily using the fastflush technique.
A fast flush is produced by opening the valve of the
continuous flush device, for example, by pulling and
quickly releasing the pigtail valve on a continuous flush
1.2
1.1
1
.9
.8
.7
.6
.5
.4
.3
.2
.1
0
DAMPED
UNACCEPTABLE
= 0.1
500
571
AMPLITUDE RATIO
.1
.2
.3
.4
.5
.6
.7
.8
.9
OPTIMAL
ADEQUATE
10
15 20 25 30 35
FREQUENCY (HZ)
40
45
50
Figure 6. Plot of fn versus z illustrating the five areas into which

catheter-tubing-transducer systems fall. Systems in the optimal
area will reproduce even the most demanding (fast heart rate and
rapid systolic upstroke) arterial or pulmonary artery waveforms
without distortion. Systems in the adequate area will reproduce
most typical patient waveforms with little or no distortion. All
other areas will cause serious and clinically important waveform
distortion. Note the scale on the right can be used to estimate z
from the amplitude ratio determined during fast flush testing (11).
See Fig. 8 for an example of waveforms. [Reproduced from Ref. 6,
with permission.]
device. The rapid valve closure generates a near square

wave pressure signal from which fn and of the cathetertubing-transducer system can be measured.
Once the fast-flush test has been executed two or three
times, the dynamic response characteristics (fn and) can
quickly and easily be determined. Natural frequency fn can
be estimated by measuring the period of each full oscillation on a strip-chart recorder following a fast flush (Fig. 7a)
and calculating the frequency from the period. Damping
coefficient can be determined by using the amplitudes of
any two successive peak-to-peak values measured after a
fast flush. The amplitude ratio is calculated by dividing the
measured height of the smaller peak-to-peak value by that
of the amplitude of the larger peak-to-peak value (Fig. 7b).
The amplitude ration can then be converted to a damping
coefficient by using the scale in the right side of Fig. 6.
Once fn and have been determined, these data can be
plotted on the graph of Fig. 6 to ascertain the adequacy of
dynamic response. Some bedside monitors and recorders
may compromise the clinical users ability to use the fastflush technique because the monitors have built-in lowpass filters. These filters should be expanded to at least 50
Hz or be eliminated.
Several factors lead to poor dynamic responses: (1) air
bubbles in the system usually caused by a poor initial
catheter-tubing-transducer system setup, (2) pressure tubing that is too long, too compliant, or a diameter that is too
small, and (3) pressure transducers that are too compliant.
The best way to enhance the systems dynamics is to
improve fn.
Invasive pressure monitoring systems have patient
risks, such as a source of infection and air embolism. In
addition, great care is required by clinical users to optimize
572
(a)
200
200
HR = 107
(c)
Fn = 7 Hz
Zeta = .1
200
2 mm
100
100
(a)
124 / 78
MP = 96
Fn = 10 Hz
Zeta = 3.5
(b)
200
100
141 / 76
MP = 96
Fn = 25 Hz
Zeta = .15
(d)
200
100
100
109 / 84
MP = 96
128 / 77
MP = 96
200
Fn = 10 Hz
Zeta = .5
(b)
100
100
22.5 mm
125 / 78
MP = 96
13 mm
0
Figure 7. Measuring dynamic response parameters from the fastflush waveform, (a) The natural frequency fn can be determined by
using a strip-chart recording to measure the period of one full
oscillation, as shown. In this example, one full cycle is 2 mm and
at a paper speed of 25 mm
s1 this results in fn 12.5 Hz
25 mm
s1/2 mm. (b) Determining the damping coefficient z
required measuring two successive peak-to-peak values of the
resulting oscillations. The amplitude ratio of the two successive
peaks is taken, giving a value of 0.58 13/22.5. With use of the
amplitude ratio and the scale on the right side of Fig. 6, the
damping coefficient z 0.17. Plotting the natural frequency and
damping coefficient on Fig. 6 shows that this system is
underdamped.
dynamic response and proper zeroing to provide accurate

and reliable data. Merely looking at pressure waveforms
will not provide the information required to determine the
adequacy of the systems dynamic response (see Fig. 8). Fastflush testing to determine these parameters is essential.
SIGNAL AMPLIFICATION, PROCESSING, AND DISPLAY
Once the pressure signal has been transmitted to the transducer, the bedside monitor operates on that signal. Most
monitors not only display the heart rate and systolic, diastolic, and mean pressure, but they also display the processed
waveform on an oscilloscope and provide an analog output for
a recorder or for transmission to a central display.
Placement of the Pulmonary Artery Catheter
The balloon-tipped, flow-directed, pulmonary artery catheter (SwanGanz) came into widespread use in 1970 (21).
(e)
200
Figure 8. Arterial pressure waveforms were obtained from the

same patient. Shown are Systolic/Diastolic and Mean Pressure
(MP). In panel the (a) Patients actual arterial pressure waveform
as if recorded with a catheter-tipped transducer is shown, (b)
shown the same patients arterial waveform recorded with an
overdamped system (z 3.5). Note the fast-flush signal (upper
left) returns slowly to the patient waveform. Systolic pressure is
underestimated, diastolic pressure is overestimated, and MP is
unchanged, (c) An underdamped condition (z 0.1) with low fn 7 Hz.
After the fast flush, the pressure signal oscillates rapidly (rings).
Systolic pressure is overestimated, diastolic is slightly underestimated, and MP is correct, (d) shows an underdamped condition
(z 0.15), but with high fn 25 Hz. The pressure waveform is slightly
distorted and systolic, diastolic, and mean pressures are close to the
actual pressures, (e) shown an ideally damped pressure monitoring
system (z 0.5). The undershoot after the fast flush is small and the
original patient waveform is adequately reproduced. [Reproduced
from Ref. 6, with permission.]
The follow-up development by Ganz of a practical thermal

dilution attachment to the pulmonary artery catheter
permitted convenient and easy measurement of cardiac
output (22). Since these early developments with the
SwanGanz catheter, the pulmonary artery catheter has
been fitted with optical fibers which allow measurement of
mixed venous oxygen saturation (23).
The pulmonary artery catheter is inserted into the right
side of the circulation using the percutaneous technique
typically using entry from either the internal jugular or the
subclavian vein. The catheter is floated into the pulmonary
artery without use of fluoroscopy, using the hemodynamic
pressure waveforms as a guide (Fig. 9).
Accurate Measurement of Pulmonary Artery Pressure.
Since it was introduced, the balloon-tipped, flow-directed,
pulmonary artery catheter (SwanGanz) has been widely
used in intensive care units. The ease with which it is
573
Figure 9. Composite illustration showing normal pressure waveforms obtained as a fiber optic
balloon flotation pulmonary artery catheter (Swan-Ganz) is advanced from the right atrium to the
pulmonary artery wedge position. [From Daily and Tilkian in Reading List (1986), with permission.]
usually inserted may lead one to conclude that the measurements of pulmonary artery and wedge pressure (PCWP) are
easily and reliably measured. However, such is not the case.
Pulmonary artery pressures can be measured accurately only if the following steps are taken (2427):
1. The monitor is properly zeroed.
2. Strip-chart recordings of all PA pressures for a time
period covering at least three respiratory cycles are
obtained. Using only the monitors digital displays is
insufficient.
3. Dynamic response testing (fast flush) should be
obtained when the catheter is in each position (i.e.,
wedge and PA). If the dynamic response is not
adequate, the problems with the catheter-tubingtransducer system must be resolved before accurate
pressures can be measured.
4. Pressures (i.e., systolic, diastolic, and mean pressures) should be assessed from a monitors display
or a strip-chart recording. The pressure measures
should be made at the end expiration when the
transmural pressure is nearest zero.
CARDIAC OUTPUT DETERMINATION
Cardiac output is the volume of blood ejected by the heart
every minute. Cardiac output is a helpful measurement
since it can be used to evaluate the overall cardiac status of
the critically ill patient, as well as help make the diagnosis
of cardiovascular disease. Ideally a cardiac output

measurement system would be continuous, automatic,
minimally invasive, accurate, fast, inexpensive, and easy
to use clinically. The most common method used to measure cardiac output in critically ill patients is still the
indicator dilution method. The pulmonary artery catheter
(SwanGanz) introduced in the 1970s revolutionized the
ease with which cardiac output could be measured.
The thermal dilution method requires injection of cold
physiological solution, usually normal saline, into the
superior vena cava or right atrium. Cardiac output is
determined by measuring the area under the time
temperature curve measured in the pulmonary artery
that results from the injection of the cold solution.
The thermal dilution method for determining cardiac
output relies on several assumptions that are not always
correct. First, the exact amount of thermal indicator
injected cannot be quantitated precisely. Second, indicator
is lost at various stages and this loss of indicator (heat loss)
leads to errors.
A block diagram of the thermal dilution measuring
system with typical thermal dilution curves and time of
injection indicated are shown in Fig. 10. Figure 10c and d
show the transit time for the cooled blood moving from the
injection site in the right atrium to the pulmonary artery
measurement site. Calculation of cardiac output requires
measuring the area under the curve. Consequently, a
baseline temperature must be established before the injection. In turn, the end point is usually determined by
extrapolating to the baseline temperature.
574
Figure 10. Schematic diagram of the

thermal dilution measurement of cardiac
output. A recorder of some type should
always be used to verify the quality of
the thermal dilution curve, (a) shows the
thermal dilution catheter placed into the
pulmonary artery. Note the location of
injection site and thermistor, (b) shows the
connection of the thermal dilution catheter
connected to a cardiac output processor and
recorder, (c) shows a typical temperaturetime plot sensed by the thermistor near the
catheter tip after an iced saline injection.
The cardiac output determined in this case
was 4.36 L
min1. (d) Shows a similar
temperaturetime plot for a patient with
low cardiac output (2.18 L
min1). Note
the larger area and broader dispersion of
the waveform caused by the lower flow.
[Reproduced from Ref. 9, with permission.]
To ensure that accurate thermal dilution cardiac output

results are obtained, it is recommended that the thermal
dilution curves be presented on a monitoring screen or on a
strip-chart recorder. Studies have shown that synchronizing the injections with the respiratory cycle improves the
techniques reproducibility (28). Since there is considerable
variability in cardiac output between measures, at least
three reproducible curves are usually obtained. Averaging
the findings from these three curves gives a more representative assessment of cardiac output.
In recent times, the complexities of using the Swan
Ganz pulmonary artery catheter have result in controversies. Some clinicians feel that such systems should only be
used only when needed and then only sparingly while
others have a differing viewpoint (29,30). Still others have
questioned the ability of making accurate central venous
and pulmonary artery occlusion pressures and whether it
matters (26,27). Many of those issues will be resolved in the
future when there might be better methods for measuring
hemodynamic parameters. Until that time, physicians and
nurses caring for critically ill patients who require hemodynamic monitoring should be aware of several how to
oriented publications (3134).
Alarming Based on Hemodynamic Parameters. Clinical
hemodynamic monitoring is now several decades old.
What started from a simple beginning has since seen

many dramatic changes in both the development of new
medical devices and skills of the clinicians to use those
devices. However, it is my feeling that we are not yet at
optimum hemodynamic monitoring. Some recent publications on the topic are illustrative. Sander and colleagues in Germany have recently looked at categories of
patients with elevated heart rates who are at higher risk
of cardiac complications (35). Their work resulted in an
editorial comment Vital are vital signs (36). Additional
recent work at Vanderbilt University indicates that
volatility in vital signs is a good predictor of death in
trauma patients (37). Finally, the problem of false
alarms continues to be a huge problem with current
bedside monitors. As part of a Master of Science thesis
in Medical Informatics at the University of Utah, an
investigator found that only about one-third of the standard alarms for patients in a variety of ICU care were
true alarms. Thus about two-thirds of the alarms are
false. However, if the alarming system used heart rates
determined from both the ECG and the Arterial Blood
Pressure, the number of alarms decreased by 50% and
the false alarm rate was only 25% (38). Having smarter and better hemodynamic monitoring with better and
smarter alarming systems will be crucial for to future
monitoring systems.
COMPUTERIZED DECISION SUPPORT

Much has been learned about hemodynamic measurements and how to use the data to calculate derived patient
parameters. These parameters can then be used to
determine patient status and augment patient therapy
(3942).
Using hemodynamic data available from bedside monitors and combining that data, in a structured and coded
electronic patient record allows for optimal computerized
decision support (4244). Morris and his colleagues have
stated the value of computerized decision support well (45).
Only adequately explicit protocols contain enough detail to
lead different clinicians to the same decision when faced
with the same clinical scenario. Guidelines of care provide
only general guidance to patient care and require clinicians
considerable latitude in which care decision should be
made. Computerized protocols, on the other hand, can be
patient-specific and evidence based (46). Using computerized decision-support tools variation in clinical practice can
be reduced and favorable effects on improve patient outcomes can be accomplished (45,46).
FUTURE
There are still needs for improvement in hemodynamic
monitoring. Being able to make the measurements continuously, less invasively, and more reliably are areas
where progress is needed. Clearly, using computer aided
decision-support technology to help reduce false alarms
and to guide clinicians in making better patient diagnosis
and more timely and more optimal and effective treatment
decision offer ample opportunity for future research and
progress.
BIBLIOGRAPHY
1. Gravenstein JS, Paulus DA. Monitoring Practice in Clinical
Anesthesia. Philadelphia (PA): Lippincott; 1982.
2. Bruner JMR. Handbook of Blood Pressure Monitoring.
Littleton (MA): PSG Publishing Co.; 1978.
3. Daily EK, Schroeder JS. Techniques in Bedside Hemodynamic Monitoring. 3rd ed. St. Louis (MO): Mosby; 1985.
4. Pinsky MF. Functional hemodynamic monitoring. Intensive
Care Med 2002;28:386388.
5. Pinsky MF. Hemodynamic monitoring in the intensive care
unit. Clin Chest Med 2003;24:549560.
6. Gardner RM, Hollingsworth KW. Optimizing ECG and pressure monitoring. Crit Care Med 1986;14:651658.
7. Pickering TG, et al. Recommendations for blood pressure measurement in humans and experimental animals: part 1: Blood
pressure measurement in humans: A statement for professionals from the subcommittee of professional and public education of the American Heart Association council on high blood
pressure research. Circulation 2005;111:697716.
8. Gardner RM. Hemodynamic monitoring: From catheter to
display. Acute Care 1986;12:333.
9. Gardner RM, Schwartz R, Wong HC, Burke JP. Percutaneous
indwelling radial-artery catheters for monitoring cardiovascular function (Prospective Study of the Risk of Thrombosis
and Infection). N Engl J Med 1974;290:12271231.
575
10. Kline AM. Pediatric catheter-related bloodstream infections:

Latest strategies to decrease risk. AACN Clin Issues 2005;16:
185198.
11. Bone RC, et al. Standards of evidence for the safety and
effectiveness of critical care monitoring devices and related
interventions. Coalition for critical care excellence: Consensus Conference on Physiological Monitoring Devices. Crit
Care Med 1995;23:17561763.
12. Kutik MH, Gardner RM. Standard for Interchangeability
and Performance of Resistive Bridge Blood Pressure Transducers, Arlington (VA): Association for the Advancement of
Medical Instrumentation (AAMI); 1986.
13. Gardner RM. Accuracy and reliability of disposable pressure
transducers coupled with modern pressure monitors. Crit
Care Med 1996;24:879882.
14. McCann II UG, et al. Invasive arterial BP monitoring in
trauma and critical care. Effects of variable transducer level,
catheter access, and patient position. Chest 2001; 120:1322
1326.
15. Gardner RM. Direct blood pressure measurement dynamic
response requirements. Anesthesiology 1981;54(3):227236.
16. Gardner RM. Blood pressure monitoring. In: Webb AR,
Shapiro MJ, Singer M, Suter PM, editors. Oxford Textbook
of Critical Care. Oxford University Press. 1998; p 10871090.
chapt 16.
17. Gardner RM. Fidelity of recording: Improving the signal-tonoise ratio. In: Martin J. Tobin, editors. Principles and Practice
of Intensive Care Monitoring. New York: McGraw-Hill; 1997.
p. 123132. chapt. 8.
18. Kleinman B, Powell S, Kumar P, Gardner RM. The fast flush
test measures the dynamic response of the entire blood pressure
monitoring system. Anesthesiology 1992;77:12151220.
19. Kleinman B, Powell S, Gardner RM. Equivalence of fast flush
and square wave testing of blood pressure monitoring
systems. J Clin Monit 1996;12(2):149154.
20. Promonet C, et al. Time-dependent pressure distortion in a
catheter-transducer system: Correction by fast flush.
21. Swan HJC, et al. Catheterization of the heart in man with the
use of a flow directed balloon-tipped catheter. N Engl J Med
1970;283:447451.
22. Ganz W, et al. A new technique for measurement of cardiac
output by thermodilution in man. Am J Cardiol 1971;27:392
396.
23. Cole J, Martin WE, Cheung PW, Johnson CC. Clinical studies
with a solid state fiberoptic oximeter. Am J Cardiol 1972;29:
383388.
24. Morris AH, Chapman RH, Gardner RM. Frequency of wedge
pressure errors in the ICU. Crit Care Med 1985;13:705708.
25. Cengiz M, Crapo RO, Gardner RM. The effect of ventilation
on the accuracy of pulmonary artery and wedge pressure
measurements. Crit Care Med 1983;11:502507.
26. Rizvi K, et al. Effect of airway pressure display on interobserver agreement in the assessment of vascular pressure in
patients with acute lung injury and acute respiratory distress
syndrome. Crit Care Med 2005;33:98103.
27. Liebowitz AB. More reliable determination of central venous
and pulmonary artery occlusion pressures: Does it matter?
Editorial Crit Care Med 2005;33:243244.
28. Stevens JH, et al. Thermodilution cardiac output measurements: Effects of the respiratory cycle on its reproducibility.
J Am Med Assoc 1985;253:22402242.
29. Pinsky MR, Vincent J-L. Lets use the pulmonary artery
catheter correctly and only when we need it. Point of view.
Crit Care Med 2005;33:11191122.
30. Levin PD, Sprung SL. Another point of view: No Swan song for
the pulmonary artery catheter. Crit Care Med 33:11231124.
576
31. Gibbs NC, Gardner RM. Dynamics of invasive pressure

monitoring systems: Clinical and laboratory evaluation.
Heart Lung 1988;17:4351.
32. Gardner RM, Hujcs M. Fundamentals of physiologic monitoring. AACN Clinical Issues Crit Care Nursing 1993;4(1):1124.
33. Daily EK. Hemodynamic waveform analysis. J Cardiovasc
Nurs 2001;15(2):622.
34. McGhee H, Bridges EJ. Monitoring arterial blood pressure:
What you may not know. Crit Care Nurse 2002;22:6078.
35. Sander O, et al. Impact of prolonged elevated heart rate on
incidence of major cardiac events in critically ill patients with
a high risk of cardiac complications. Crit Care Med 2003;
33:8188.
36. Weissman C, Landesberg G. Vital are the vital signs. Comment/Editorial Crit Care Med 2003;33:241242.
37. Grogan EL, et al. Volatility: A new vital sign identified using
a novel bedside monitoring strategy. J Trauma 2005;58:714.
38. Poon KB. Fusing Multiple Heart Rate Signals to Reduce
Alarms in Adult the Intensive Care Unit. MS dissatation,
University of Utah Department of Medical Informatics, Salt
Lake City (UT); May 2005.
39. Gardner RM. Information management hemodynamic monitoring. Semin Anesth 1983;2:287299.
40. Gardner RM. Computerized management of intensive care
patients. MD Comput 1986;3(1):3651.
41. Shabot MM, Carlton PD, Sadoff S, Nolan-Avila L. Graphical
reports and displays for complex ICU data: A new, flexible
and configurable method. Comput Methods Programs
Biomed 1986;22:111116.
42. Gardner RM, Huff SM. Computers in the ICU: Why? What?
So what? Intl J Clin Monit Comput 1992;9:199205.
43. Gardner RM, Sittig DF, Clemmer TP. Computer in the ICU:
A match meant to be! In: Ayers SM, Grenvik A, Holbrook PR,
Shoemaker WC. editors. Textbook of Critical Care Medicine.
3rd ed. Philadelphia: W. B. Saunders;1995: p 17571770.
chapt. 196.
44. Morris AH. Treatment algorithms and protocolized care.
Curr Opin Crit Care 2003;9:236240.
45. Morris AH. Clinical trial of a weaning protocol. Crit Care
2004;8:207-209.
46. Bria II WF, Shabot MM. The electronic medical record, safety
and critical care. Crit Care Clin 2005;21:5579.
Further Reading
Nichols WW, ORourke MF. McDonalds Blood Flow in Arteries:
Theoretical, Experimental and Clinical Principles. 5th ed. Of
special note is Chapter 6: Measuring principles of arterial
waves pages 132135. Hodder Arnold; 2005.
Daily EK, Tilkian AG. Hemodynamic Monitoring. In: Tilkian AG,
Daily EK, editors. Cardiovascular Procedures: Diagnostic
Techniques and Therapeutic Procedures. St. Louis (MO):
Mosby; 1986 chapt. 4.
Geddes LA. The Direct and Indirect Measurement of Blood
Pressure. Chicago: Year Book Medical Publisher; 1970.
Geddes LA. Handbook of Blood Pressure Measurement. Clifton
(NJ): Humana Press; 1991.
Webster JG, editor. Design of Pulse Oximeters. Institute of Physics; 1997.
Webster JG, editor. Medical Instrumentation: Application and
Design. 3rd ed. New York: John Wiley & Sons Inc.; 1998.
Tungjitkusolmun S, Heart and Circulation, In: Webster JG, editor.
Bioinstrumentation. John Wiley & Sons, Inc.; 2004. Chapt. 8.
Shabot MM, Gardner RM, editors. Decision Support Systems for
Critical Care. New York: Springer-Verlag Inc.; 1994.
Gardner RM, Shabot MM. Patient-monitoring systems. In: Shortliffe EH, Cimino JJ, editors. Biomedical Informatics: Computer
Applications in Health Care. 3rd ed. New York: SpringerVerlag; 2005; in press, Aug. 2005. Chapt. 17.
See also BLOOD
PRESSURE MEASUREMENT; CARDIAC OUTPUT, FICK TECH-
NIQUE FOR; CARDIAC OUTPUT, INDICATOR DILUTION MEASUREMENT; CARDIAC

OUTPUT, THERMODILUTION MEASUREMENT; ELECTROCARDIOGRAPHY, COMPUTERS IN; HEMODYNAMICS.

MICHAEL L. DALEY
IAN PIPER
The University of Memphis
Memphis, Tennessee
INTRODUCTION
Intracranial pressure (ICP) monitoring is a form of pressure monitoring used in patients with pathologies that
might give rise to raised ICP. The definition of raised ICP
depends on the underlying pathology and, for example in
adult patients who have sustained a severe head, injury is
defined as pressure greater than 20 mm Hg. Over the past
50 years there has been an active and wide ranging
research into the causes and consequences of raised
ICP that, to date, has been the subject of 11 international
symposia embracing such diverse disciplines as neurosurgery, anesthesia, radiology, biophysics, electrical and
mechanical engineering, mathematics, and computer
science. This article reviews the underlying physiology
pertinent to measurement of ICP as well as gives an
overview of some of the current technology used to measure ICP. Then, a brief review follows of the clinical
literature underlying the case for ICP measurement, with
an emphasis on the main clinical condition of patients
with a head injury. The last two sections focus on the use
of waveform analysis and mathematical modeling techniques to research into mechanisms underlying raised
ICP.
PHYSIOLOGY
ICP is the pressure recorded within the tissue (parenchyma)
or fluid-filled spaces and is not uniformly distributed within
the craniospinal axis. The craniospinal axis consists of all
neural tissue and contiguous fluid-filled spaces within the
cranium and spinal sac. The central neural tissue is encapsulated within bone and three-layered tissue coverings or
meninges. From outside in the meninges are the dura,
arachnoid, and pia membranes. The meninges provide a
physical barrier between neural tissue and the external
environment but also serve both a structural and a physiological function as an architecture for supporting cerebral
vessels and maintaining a space for flow of cerebrospinal
fluid (CSF). CSF cushions the delicate neural tissues, supports the weight of the brain, and acts as a transport media
for nutrients, chemical messengers, and waste products.
Except at the choroid plexus, the CSF is freely permeable
to the appendymal lining and is in chemical communication

with the interstitial fluid and the CNS. ICP is a reflection of
the relationship between alterations in craniospinal volume
and the ability of the craniospinal axis to accommodate
added volume. The craniospinal axis is essentially a partially closed box with container properties including both
viscous and elastic elements. The elastic or, its inverse, the
compliant properties of the container will determine what
added volume can be absorbed before ICP begins to rise. So
an understanding of raised ICP encompasses an analysis of
both intracranial volume and craniospinal compliance.
The history of the subject of ICP has been well reviewed
(1,2) and starts from the doctrine named after Monro (3)
and Kellie (4), which proposed that the brain and its
contained blood were incompressible, enclosed in a rigid
and inextensible skull, of which the total volume remained
constant. In its original form, the MonroKellie doctrine
did not take into account the CSF as a component of the
cranial compartment. The concept of reciprocal volume
changes between blood and CSF was introduced in 1846
by Burrows and later extended in the early twentieth
century by Weed et al. (5,6) to allow for reciprocal changes
in all craniospinal constituents. Kocher (7) in 1901 translated into clinical terms the four stages of cerebral compression proposed almost 25 years earlier by the
experimental studies of Duret (8). Kocher described four
stages of cerebral compression related to the expansion of
intracranial brain tumours. In stage 1, the initial increase
in tumor volume is compensated by a reduction in volume
of the other intracranial components, chiefly CSF and
venous blood. This spatial compensation results in no
net increase in intracranial volume or pressure and hence
no clinical symptoms. In stage 2, the compensatory
mechanisms are exhausted, ICP increases, and the patient
becomes drowsy with a headache. Stage 3 is characterized
by a considerable increase in ICP, an associated deterioration in conscious level, and intermittent elevations of blood
pressure (BP) accompanied by bradycardia. In the fourth
and final stage, the patient is unconscious, with bilateral
fixed dilated pupils and falling BP usually leading to death.
Cushing (911), then a research worker for Kocher,
described in both experimental and clinical studies the
close relationship between increases in ICP and BP and
proposed that the BP rose to maintain adequate blood
supply to the hind brain, the stimulus to this vasopressor
response believed to be medullary ischemia (12,13). At
about this time, a false confidence developed in the lumbar
CSF pressure technique (lumbar puncture), which caused
Cushings findings to be challenged. Reports emerged
(1416) that some patients showing clinical signs of brain
compression had normal lumbar CSF pressures and that
in other patients elevations in BP were found at times
when ICP was well below the level of BP.
Partly because of this apparent dissociation between
ICP and clinical symptoms, emphasis switched away from
ICP measurement toward the relationship between craniospinal volume and pressure, particularly the importance of the elastic properties of the craniospinal system.
The relationship between ICP and intracranial volume is
often described graphically by an exponential function that
is relatively flat at low ICPs but rises rapidly as pressure
577
increases much above 20 mm Hg. Under normal conditions,

at the low ICP end of this curve, compliance (the ratio of
added volume to pressure) is high. When the patients
volumepressure status changes to move further up the
curve, compliance will fall rapidly. Measurement of craniospinal compliance in brain-injured patients may, therefore, offer the potential for early detection of raised ICP
before it rises to levels damaging to brain parenchyma. The
most commonly used methods of measuring craniospinal
compliance were developed by Marmarou (17,18) and
depend on the rapid injection of known volumes of fluid
into the CSF space with immediate measurement of the
resultant increase in CSF pressure. One of these methods
is the pressure volume index (PVI); this value being the
volume that when added to the CSF space would produce a
tenfold rise in ICP. Miller et al. (19,20) defined a further
measure of the craniospinal volumepressure relationship,
the volume pressure response (VPR). The VPR, also calculated from the ICP response resulting from a rapid bolus
injection of saline into the CSF space, is a direct measure of
the inverse of compliance: elastance. Although both are
measures of compliance, some confusion remains as to
what exactly is the difference between the VPR and the
PVI. The PVI, which assumes a mono-exponential pressure
versus volume relationship, is a single index that characterizes the patients entire volumepressure relationship
and is, numerically, a measure of the slope of the logarithm
of the ICP versus intracranial volume relationship. Its
strength lies in its ability to define the whole volume
pressure status of the patient with a single index. It was
the late J. Douglas Miller who pointed out that if there
was only a single volumepressure curve, then no new
information would be gained by measuring compliance or
elastance, and a knowledge of absolute ICP alone would
suffice in determining the state of a patients craniospinal
volume decompensation. However, several studies (21,22)
have shown that the shape of the volumepressure relationship changes under a variety of conditions both
between patients and within patients at different times.
Under these circumstances, it is likely that the PVI will
provide the more useful information. One weakness of the
PVI is that if a patients pressurevolume relationship
remains stable, single measurements will not detect
movement along a given pressurevolume curve and thus
may not detect slow increases in volume of a space-occupying
lesion before it causes significant increases in ICP. It is in
this situation where a continuous measure of absolute
compliance (or its inverse: elastance as measured by the
VPR) provides, potentially, more useful information. It is
for this reason that, in head-injured patients, Miller
et al. found that the VPR correlated better to the degree
of brain mid-line shift, as imaged on CT scan, than it did
to absolute ICP alone. They subsequently demonstrated
that the VPR could serve as an indicator for surgical
decompression, critical levels being between 3 and 5 mm
Hg/mL (19,23). However, in clinical practice, both circumstances (movement along the pressurevolume curve and
shift of the curve along the pressure axis) may occur within
a patient at different times and thus shows the need for
ICP monitoring capable of both continuous compliance
measurement and derivation of the PVI.
578
Few clinicians would argue over the theoretical utility of

monitoring compliance in brain-injured patients; what has
limited its use in practice has been the inherent problems
associated with manual volumepressure testing. It is
difficult to inject equal volumes of fluid manually at a
constant and rapid rate of injection. As a result, repeated
measures are usually required that can result in a lengthy
and time-consuming procedure. Also, even with stringent
maintenance of the sterile procedure, repeated addition or
removal of CSF from patients carries an increased risk of
infection to the patient. Despite the reported benefits of
compliance measurement, it has been largely these limitations that have prevented widespread adoption of compliance measurement in clinical practice.
It was not until the 1960s when Lundberg (24) published
his now classic monograph that interest in clinical ICP
measurement was rekindled. Using ventricular fluid
pressure recording in brain tumor patients, Lundberg
was the first to delineate the frequency with which raised
ICP occurs clinically, at times reaching pressures as high
as 100 mm Hg. Lundberg also described three types of
spontaneous pressure wave fluctuations: A waves or
plateau waves of large amplitude (50100 mm Hg) with
a variable duration (520 min), B waves that are smaller
(up to 50 mm Hg), and sharper waves with a dominant
frequency of 0.52 cycles per min.
DEVICES
ICP was first measured experimentally in animals by
Baylis (25) in 1897 using an early form of strain gauge.
The most common form of strain gauge use specially prepared alloys that change their resistance in proportion to
the amount they are elongated or stretched in cross section
due to applied strain.
The Wheatstone Bridge and the Strain Gauge
Most common strain gauge transducers are used in a
Wheatstone Bridge configuration, where the resistive
strain elements are placed on diagonally opposite arms
of the bridge. Should the strain gauge change its resistance
(for example, due to an applied strainperhaps caused by
a pressure acting on the strain gauge to cause it to elongate), then the bridge will become unbalanced and a
potential difference will be generated in proportion to
the degree of strain.
Fluid-Filled Catheter Transducer Systems
The standard intraventricular catheter connected to an
external strain gauge transducer is called a cathetertransducer system because it behaves, in many ways, like
a mechanical system with a mass of fluid that acts against
the spring-like elastic properties of the catheter walls
and the transducer diaphragm. A typical catheter-transducer system is one used for the measurement of arterial
pressure comprising a silicon strain gauge, fluid-filled
catheter, three-way tap, and an arterial cannulae. Modern
transducers are semiconductor fabricated and normally
disposable. If a flushing device is attached to the transdu-
cer, extreme care must be used to avoid accidental flushing

into the cranium. The older catheter-transducer system
included long tubing and therefore had different frequency
characteristics than modern transducers. The older transducer first found widespread use in the 1960s and 1970s
following the pioneering work on long-term ICP monitoring
by Nils Lundberg (24). Most publications in this field still
refer to intraventricular monitoring as the Gold-Standard
method of measuring ICP for several reasons. First, this
method allows checking for zero and sensitivity drift of the
measurement system in vivo. Second, pressure measurement within the CSF space transduces pressure within a
medium that is an incompressible fluid and, provided CSF
flow is not blocked, is not subject to the development of
intracompartmental pressure gradients. Finally, access to
the CSF space provides a method for ICP treatment via CSF
drainage. However, concerns are often expressed about the
increased risks of infection associated with ventriculostomy.
Although a range of infection rates has been reported, some
as high as 40% (26), recent reports confirm infection rates to
be in the region of 1%, which is not considered a prohibitive
risk (27).
The Head Injury Management Guidelines published by
the Brain Trauma Foundation in 1994 recommend intraventricular ICP measurement as the first-line approach to
monitoring ICP. They state that A ventricular catheter
connected to an external strain gauge transducer or catheter tip pressure transducer device is the most accurate and
reliable method of monitoring ICP and enables therapeutic
CSF drainage. Clinically significant infections or haemorrhage associated with ICP devices causing patient morbidity are rare and should not deter the decision to monitor
ICP (28). Despite the existence of these guidelines (27,28),
catheter-tip intraparenchymal pressure monitoring
remains popular, particularly in the United Kingdom, as
it does not require catheter placement in the operating
theater and thus carries significantly lower resource implications. However, the routine use of fluid-filled cathetertransducer systems is not without difficulties. A cathetertransducer system can be described as a second-order
mechanical system and, if under-damped, will oscillate
at its own natural frequency producing significant amplitude and phase distortion of the pressure signal. The
degree of distortion will depend on the damping factor
(b) of the system. For most purposes, a damping factor
of 0.64 is optimal as the amplitude error will be less than
2% for up to two thirds of the natural frequency of the
system (29). Typically, most pressure catheter-transducer
systems used in patients tend to be under-damped with a
damping factor (b) less than 0.4 (29,30). Manipulating a
catheter-transducer system from under-damped (b < 0.3)
to over-damped (b > 0.8) can cause a decrease in mean
pressure of 7 mm Hg. Commercial devices are available,
however, such as the acudynamic adjustable damping
device (31) that can alter the damping characteristics of
external strain gauge pressure transducers and bring them
within the range of optimal damping. Another problem
with fluid-filled catheter-transducer systems is correcting
for the presence of hydrostatic pressure gradients when
measuring cerebral perfusion pressure (CPP). Typically,
the external strain gauge transducer is zeroed at the same
579
level as the arterial pressure (BP) transducer, usually at

the level of the right atrium. Although the patient is
managed in the horizontal position, there is no column
of fluid between the site of ICP and BP measurement.
However, when the patient is managed with head-up tilt
and if the BP transducer is not moved to the same horizontal level as the head, a hydrostatic fluid column will be
created. This can produce a significant error between the
observed CPP and the actual CPP, which, in the worst case,
can produce an error of as much as 15 mm Hg.
The Spiegelberg ICP monitoring system (Spiegelberg
KG, Homburg, Germany) largely overcomes these problems. This system is a special case of a fluid-filled catheter-transducer system. With this device, ICP is measured
using a catheter with an air pouch balloon situated at the
tip. By maintaining a constant known volume within the
air pouch, the pressure within the air pouch balloon is
equivalent to the surrounding pressure or ICP. The internal air pouch balloon is transduced by an external strain
gauge transducer, and because the fluid used for pressure
transduction is air, the pressure error caused by an air
column is clinically insignificant. The design of this device
also allows automatic in vivo zeroing of the ICP system
and, in laboratory bench tests, showed the least zero drift
in comparison with standard catheter-tip ICP devices (32).
This system now has versions for use in epidural, subdural,
intraparenchymal, and intraventricular sites. The intraventricular catheter is a double lumen catheter that allows
access to the CSF space for drainage. The Spiegelberg system, although having many attractive features, is, as yet, a
relatively little used system outside of Germany and its longterm clinical utility and robustness require evaluation.
ent zero drift of the catheter once placed. For the Camino
system, the manufacturers specify the zero drift of the
catheter to be 2 mm Hg for the first day and 1 mm
Hg per day thereafter. Czosnyka et al. (32) have confirmed
these zero drift findings in bench tests studies although
they also reported that the temperature drift of the device
was significant (0.3 mm Hg/8C). They reported that if
the manufacturers specify the zero drift of the catheter
to be 2 mm Hg for the first day and 1 mm Hg per day
thereafter. In clinical practice, the reported zero drift upon
removal of the Camino device from the patient has been
reported to be greater than the manufacturers specifications. Munch (38) assessed 136 Camino sensors in a clinical
study and found an average daily drift rate of 3.2 mm Hg.
Chambers (39), in a comparative study of the Camino
ventricular catheter with an external fluid-filled cathetertransducer system, reported that only 60% of the readings
were within 2 mm Hg of the gold-standard method. There
are also reports of Camino probe failure because of technical
complications (cable kinking, probe dislocation), with
reported failure rates ranging from 10% to 25% (39,40).
The Codman transducer is a micro-miniature strain
gauge within a titanium housing side mounted at the tip
of a catheter. Similar to other transducers, bench test
reports on this technology have been favorable (32,41).
However, clinical evaluations have reported the presence
of inter-patient and intra-patient biases that are independent on whether the device is compared against the
Camino transducer or an intraventricular catheter-transducer system (42). A report by Fernades (43) found that in
24% of the recordings, the Codman sensor over-read the
Camino system by 5 mm Hg or more.
Catheter-Tip Transducer Systems
The Rehau System
Now several catheter-tip ICP monitoring systems are

available, including the Camino (3336) systems. The
Gaeltec ICP/B solid-state miniature ICP transducers, for
use in the epidural space, are reusable, and the zero reference
can be checked in vivo. However, reports of measurement
artifacts (37) and decay in measurement quality associated
with repeated use (35) have limited the widespread adoption
of this technology.
The InnerSpace OPX 100 system (InnerSpace Medical,
Irvine, CA) is, like the Camino system, a fiber optic system.
Bench test reports on this system show it to have good zero
drift and sensitivity stability (32). However, a recent clinical evaluation of this system in 51 patients reported a high
(17%) incidence of hematoma formation around the ICP
sensor (36). The authors concluded that improved fixation
of the catheter is required to minimize micro-movements.
The two catheter-tip systems most frequently used in
the management of head-injured patients are the Codman
and Camino systems. Neither allows a pressure calibration
to be performed in vivo. After these systems are zeroed,
relative to atmospheric pressure during a pre-insertion
calibration, their pressure output is dependent on zero
drift of the sensor. For this reason, it is critical that these
devices exhibit good long-term zero drift characteristics.
These devices provide an electrical calibration, to calibrate
external monitors, but they cannot be corrected for inher-
The technology of miniaturization is allowing the development of catheter-tip pressure sensors. Catheter-tip systems, due to their small diameter, are likely to cause
less damage to tissue upon placement than larger fluidfilled catheters and are not affected by hydrostatic pressure
differences. However, they are potentially more prone to
problems of robustness and in vivo zero drift. Several
similar technologies, although performing well in bench
test studies, have been shown to exhibit unacceptable zero
drift, fragility, or both during trials conducted under
clinical conditions (38,4345). In vivo drift is especially
important in catheter-tip strain-gauge technology as it is
impossible to check if the calibration has altered after being
placed in the patient. Until recently, to reduce the physical
size of the catheter-tip strain gauge catheters, often only a
partial Wheatstone bridge is employed at the catheter-tip.
Recently, a new technology has become available, the
Neurovent-P (REHAU AGCO, REHAU, Germany), in
which a full Wheatstone bridge is fabricated in the probe
tip, and this solution should, in theory, provide improved
zero drift characteristics. One potential advantage of the
Rehau NeuroVent system seems to be the incorporation of
a full Wheatstone bridge into the catheter-tip electronics. This
Wheatstone technological improvement should enhance
the zero drift characteristics by reducing temperature
sensitivity and the effects of non-pressure-related external
580
strains. Until recently, this approach was only possible

in the much physically larger external strain gauge systems. Through advances in miniaturization technology, it
has now been able to incorporate this technology into its
catheter-tip systems. In the bench studies performed (46),
each catheter was tested for days, from a minimum of 3 to
more than 8 days. This timeframe typically covers the
period in which an ICP transducer is used in the clinical
setting. The results from this bench test study confirm the
manufactures claim that pre-insertion calibration is not
required as the drift was within manufactures specifications (1 mm Hg). This advantage is very important in the
clinical arena because the first pre-insertion zeroing is a
crucial step conducted by the surgeons and potentially, if
incorrectly performed, could generate erroneous readings
during the period of monitoring. Both long-term zero drift
and the dynamic pressure test results also confirm that
this system performs well in bench test studies and
meets manufacturers specifications. Mean zero drift, after
5 days, was very small and long-term continuous recording
of a stable pressure was precise. The response to dynamic
tests, i.e., the changes of pressure over a wide range, was
excellent. The average bias of the Rehau catheter compared with a hydrostatic pressure column is very small.
Despite these promising bench test study results, further
work is required to determine the performance of this
measurement device in the clinical environment. Following on from these bench tests, the next and most critical
step will be to conduct a trial of this promising technology
under the more demanding clinical environment. The
BrainIT group (http://www.brainit.org) as a multicenter
collaborative group of neurointensive care scientists and
clinicians are well placed to design and conduct such a
trial (47). Should this technology demonstrate, under clinical conditions, the required robustness and low drift as
indicated by these bench test studies, it may lead to more
precise and reliable measurement of ICP.
CLINICAL LITERATURE
Raised ICP has been found to be associated with a poorer
outcome from injury with the higher the level of ICP,
particularly the peak ICP level, which has been found to
correlate with the expected prognosis for mortality and
morbidity (4851). There has, however, been controversy
over the usefulness of monitoring raised ICP with some
groups, with a no ICP monitoring policy, finding in their
studies of head injury mortality and morbidity that outcome is similar to other groups that do monitor ICP (52).
Reported differences in the utility of ICP monitoring could
be due to variability both in management and in monitoring protocols between different neurosurgical centers. Variation in type of ICP pressure monitor, site of placement,
treatment thresholds, patient referral characteristics, and
in outcome measures can all combine to produce a large
variability both in measured ICP and in outcome irrespective of whether ICP is monitored or how it is treated.
Another source of variation in terms of raised ICP is the
inherent variability of the head-injured population with
outcome being dependant on several other factors. For
example, mass lesions are generally accompanied by elevations in ICP of greater than 40 mm Hg and are associated
with poorer outcome, whereas diffuse injuries tend to have
lower ICP levels associated with a similar poor outcome
(50,53). Age is also an important factor with an age-dependent distribution of ICP for both type of injury and outcome. This is particularly so for pediatric cases (5456).
ICP can even be raised in the absence of overt signs of
swelling or mass lesions on CT. In a small study of severely
head-injured patients, OSullivan (57) demonstrated that
some comatose head-injured patients whose initial CT scan
was normal, with no mass lesion, midline shift, or abnormal basal cisterns, developed raised ICP greater than 20
mm Hg that lasted longer than 5 min. This included a
subset of patients showing pronounced raised ICP of
greater than 30 mm Hg.
Data from large prospective trials carried out from
single centers and from well-controlled multicenter studies
have provided the most convincing evidence for a direct
relationship between ICP and outcome (5861). Narayan et
al. (58) in a prospective study in 133 severely head-injured
patients demonstrated that the outcome prediction rate
was increased when the standard clinical data such as age,
Glasgow Coma Score on admission (GCS), and pupillary
response with extraocular and motor activity was combined with ICP monitoring data. Marmarou et al. (60),
reporting on 428 patients data from the National Institute
of Healths Traumatic Coma Data Bank, showed that
following the usual clinical signs of age, admission motor
score, and abnormal pupils, the proportion of hourly ICP
recordings greater than 20 mm Hg was the next most significant predictor of outcome. Outcome was classified by
the Glasgow Outcome Score (GOS) at 6 months follow-up.
They also found, using step-wise logistic regression, that
after ICP, arterial pressure below 80 mm Hg was also a
significant predictor of outcome. Jones et al. (61) studied
prospectively 124 adult head-injured patients during
intensive care using a computerized data collection system
capable of minute by minute monitoring of up to 14 clinically
indicated physiological variables. They found that ICP,
above 30 mm Hg, arterial pressure below 90 mm Hg, and
cerebral perfusion pressure below 50 mm Hg significantly
affected patient morbidity.
Although differing opinions remain about the contribution of continuous monitoring of ICP to reduction in mortality and morbidity after head injury, there is now
sufficient evidence to remove doubt about the value of
ICP monitoring toward improving the detection and preventative management of secondary cerebral injury.
Raised Intracranial Pressure: Relationship to Primary
and Secondary Injury
Both experimental and clinical studies have clearly shown
that after traumatic brain injury, normal physiological
mechanisms for maintaining cerebral perfusion can
become impaired (6265). These studies demonstrate that
brain injury can cause impairment or loss of autoregulation
defined as the ability of the cerebral vessels to respond to
changes, in arterial gases or to arterial pressure. As a
result of these changes, there can, at times, be a decrease
in cerebrovascular resistance that can lead to raised ICP in

both adults and children (6670). Although brain-injured
patients are being managed in intensive care, there are,
superimposed on to the primary injury, periods of reduced
arterial PO2 or episodes of arterial hypotension often as a
result of other injuries or treatment by hypnotic drugs
(50,53,60,61,71,72). With an impaired physiological
mechanism unable to respond adequately to these adverse
changes in physiological parameters (or secondary
insults), ischemic brain damage can occur. These secondary, chiefly ischemic brain insults are common, with
Graham et al. (74) reporting, in a series of 151 fatal cases
of severe head injury, a 91% incidence of ischemic brain
damage found on autopsy. A second study carried out by
the same group over 10 years later found a similar high
incidence (>80%) of ischemic brain damage despite subsequent improvements in intensive care of head-injured
patients (74).
There has been much interest in the relationship among
ICP, CPP, and CBF. The landmark study of Miller and
Garibi (75) produced some of the first experimental evidence confirming the concept that changes in ICP affect
cerebral blood flow not directly but through changes in
CPP, where CPP is defined as the difference between mean
arterial pressure and ICP. Strictly speaking, the actual
cerebral perfusion outflow pressure would be cerebral
venous pressure, although this pressure is, in most situations, impractical to measure routinely. However, it has
been established that over a wide range of pressures
cerebral venous pressure is well approximated (within
34 mm Hg) by ICP (7678). In an experimental study
of CBF as determined by the venous outflow technique in
dogs, Miller and Garibi (75) also demonstrated that when
MAP and ICP rise in parallel so that CPP remains constant
at 60 mm Hg, CBF increases with MAP in animals found to
be non-autoregulating. It was further shown that as CPP
drops in autoregulating animals, the breakpoint at which
CBF starts to decrease is at a higher level if CPP is reduced
through hemorrhagic arterial hypotension than through
intracranial hypertension. This work suggests that cerebral perfusion is more sensitive to arterial hypotension
than to intracranial hypertension.
The clinical significance of this information is that in the
management of head injury, it is often necessary to employ
therapy to lower raised ICP. Therapeutic agents for reducing raised ICP often do so at the expense of reduced MAP,
and as a consequence, CPP may not improve. If autoregulation is preserved, CBF should remain unchanged despite
parallel changes in MAP and ICP. However, clinically,
autoregulation is likely to be impaired in those conditions
in which ICP is increased such as head injury or subarachnoid hemorrhage (68,69,7982). Under these circumstances, it is important that reduction in ICP should not be
achieved at the expense of lowering CBF and provoking
brain ischemia.
This earlier work of Miller and Garibi was later
extended by Chan et al. (83) to include CPP ranges of
60, 50, and 40 mm Hg. At CPP levels of 50 and 60 mm
Hg, when autoregulation was intact, CBF remained
unchanged. However, with loss of autoregulation, there
was a trend for CBF to increase as MAP and ICP were
581
increased in parallel at a CPP of 50 and 60 mm Hg.

Absolute CBF levels were significantly different between
the autoregulating and non-autoregulating groups. At a
CPP of 40 mm Hg CBF showed a linear correlation with BP.
This work demonstrates that when autoregulation is
impaired, there is a functional difference between autoregulating and non-autoregulating cerebral vessels despite
similar MAP and CPP, and that when autoregulation is
impaired, CBF depends more on arterial driving pressure
than on cerebral perfusion pressure.
The importance of arterial pressure as the prime factor
governing CPP-related secondary insults has been well
demonstrated by the work of Jones et al. (61), where they
carried out a prospective study over 4 years of the frequency and severity with which secondary insults occur to
head-injured patients while being managed in intensive
care. They developed a microcomputer-based data collection system that allows the acquisition of data from up to 15
monitored variables minute by minute (84). At each bed
space, data collection was under the control of a microcomputer where serial links between the patient monitors
and the microcomputer allow the controlled transfer of
multiple channels of physiological data once per minute.
The controlling software allowed medical staff to add comments to the current active computer file at any time,
precisely annotating significant events. The software performs artifact detection, calculates derived data, and highlights valid data that falls outside normal physiological
levels. Collected data were stored to disk and could be
printed either locally or remotely. Later work by Chambers
et al. (85) found that both raised ICP and lowered ABP are
major factors in producing secondary insults.
From the valid physiological data produced, manual
processing of data was used to identify secondary insults
that are defined at one of three grades of severity and that
must last for 5 min or longer to be recorded as an insult.
This permits calculation of the frequency, severity, and
total duration of insults, measured in minutes.
An analysis was made of 124 adult head-injured
patients who were monitored during intensive care using
the computerized data collection system. Information was
logged at 1 min intervals and scanned to identify insults
when values fell outside threshold limits for 5 min or
longer. Three grades of insult were defined for each variable (Table 1). The duration of insults has been analyzed in
relation to the GOS of these patients at 12 months after
injury. The monitored patients included 68 with severe
head injury (GCS 8 or less with no eye opening), 36 with
moderate head injury (GCS 912), and 20 with minor head
injury (GCS 1315 but with multiple injuries, scoring 16 or
more on the Injury Severity Scale).
Insults were found in 91% of patients at all degrees of
severity of head injury. Overall, 10% of patients had insults
that were only at the lowest grade 1 level, 31% had insults
at both grade 1 and grade 2 levels, and 50% of patients had
at least one insult at grade 3 level in addition to grades 1
and 2 insults. Overall, the majority (77%) of all insults
detected in the ITU were at grade 1 level, and these
represented 85% of the total duration in minutes of insult
measured. Differences in the duration of insult between
outcome groups 12 months post-injury were compared with
582
each grade of insult using KruskallWallis one-way analysis of variance and MannWhitney U tests. Significant
differences in the distribution of hypotensive insults were
found between the outcome grades at all levels of severity
of insult. Similar results were found for cerebral perfusion
pressure insult duration. These data confirm the important
adverse effect of even moderate reductions in arterial
pressure (systolic BP less than 90 mm Hg or mean BP less
than 70 mm Hg).
Although the occurrence and clinical significance of
severe and long-lasting secondary insults in head-injured
patients is not disputed. The incidence, severity, and duration of shorter acting minute by minute cerebral perfusion pressure insults, as defined by the Edinburgh
secondary insult detection methodology, has not been
defined outside of the Edinburgh study population. In
addition, the strong association between the occurrence
of specific insult types and the subsequent patient morbidity and mortality found by the Edinburgh study (61) needs
to be reproduced in other centers. From the Edinburgh
group, Signorini et al. (86,87) developed and validated a
model for predicting survival in head-injured patients
based on collection of simple demographic features. When
the minute by minute secondary insult data were added to
the baseline model, they found only ICP insults significantly improved the fit of the model.
Almost all of the evidence for a CPP management is
based on single-center cohort studies, often compared with
historical controls. For example, Rosner and Becker (26)
reported on clinical results of a CPP management protocol
where approximately 40% of patients received vasopressor
support. They reported a greatly improved incidence of
favorable outcome in patients GCS 7 compared with
historical controls. Despite evidence from single-center
studies as described above, a critique of the literature
for the purposes of defining head injury management
guidelines published by the Brain Trauma Foundation
in 1994 states that there is not sufficient evidence to
establish either a standard or a guideline for the management of CPP; however they indicate management of CPP
greater than 70 mm Hg as a management option (28). Since
then, Robertson et al. (88) performed one of the first
randomized controlled single-center trials of the CPP management approach. They defined two management cohorts,
one based on their normal practice of CPP management
and the other CBF, guided practice that included the
aggressive management of CPP above 70 mm Hg, together
with restricted use of significant hyperventilation. Conversely, their trial has shown that aggressive management
of CPP > 70 mm Hg, although reducing the incidence of
jugular venous desaturation < 50%, demonstrated no difference in neurological outcome possibly due to the
increased incidence of acute respiratory distress syndrome
in the CBF management group. Thus, secondary insults
are common, result in mainly ischemic brain damage, and
are a major contribution to disablement. Moreover Chambers et al. (89) has reported that the critical CPP thresholds
for insults of pediatric patients varies with age. For both
adult and pediatric patients, insults are important because
they are common and yet so potentially avoidable. Clearly a
critical challenge facing us is to develop patient monitoring
systems and protocols that will lead to rapid detection and

resolution of secondary insults. However, detection is not
enough; we need also improved and clinically proven methods of treating secondary insultsfurther evidence is
required. In Signorini et al.s article (87), they conclude
that the questions posed by such observational studies
can only be answered definitively within the context of a
randomized clinical trial. However, to design such a multicenter randomized clinical trial will require improved
standards in the monitoring and analysis of secondary
insult data. In Europe, improved standards for highresolution collection and analysis of multicenter data from
head-injured data is now being addressed by the Brain-IT
group (90). The Brain-IT group (http://www.brainit.org) is
an open consortium of clinicians and basic scientists
working toward improving the infrastructure for conducting both observational and controlled trials of medical
devices and patient management.
INTRACRANIAL PRESSURE ANALYSIS METHODS

Since the early 1990s, clinical monitoring of ICP has generally become part of the intensive care management of
patients with brain injury. Several methods of analysis
have been designed to extract pathophysiological information from the ICP and the corresponding ABP recordings.
As noted, one particular computation used in clinical practice is the determination of mean CPP, the pressure across
the brain. CPP is calculated as the difference between
mean ABP and mean ICP and is based on the assumption
that cerebral venous pressure is approximately equal to
ICP. CPP is a useful parameter because it provides some
insight as to whether the blood flow though brain capillaries is regulated. In the uninjured brain, cerebral blood
flow is regulated to match the metabolic demand of the
brain cells. Regulation of flow is primarily done by active
dilation and constriction of cerebral arterioles in response
to changes of CPP and/or biochemical vasoconstrictive or
vasodilator agents that interact with the vascular endothelium. The steady-state relationship between cerebral blood
flow and CPP is termed static pressure regulation and is
illustrated by the graphical relationship between cerebral
blood flow and CPP shown in Fig. 1.
In the steady state, cerebral blood flow is laminar and
computed as the ratio of CPP to hemodynamic resistance,
which is inversely proportional to the fourth power of the
radius of the vessel. In the autoregulatory range, the
arterialarteriolar bed actively adjusts the resistance of
its vessels by dilating when CPP decreases and constricting
when CPP increases to maintain a relatively constant
cerebral blood flow during changes in CPP. When CPP is
below the lower limit of the autoregulatory range, vessels
within the arterialarteriolar bed tend to passively vasoconstrict. When CPP is above the upper limit of autoregulation, passive vasodilation occurs. One proposed intensive
care therapeutic procedure designed to prevent secondary
complications during recovery is CPP-oriented therapy
(91). This therapy requires pressure autoregulation and
the ability to manipulate CPP within the autoregulatory
range (91). During intact pressure regulation, increases of
Figure 1. Illustration of static pressure regulation. Within the

autoregulatory range, changes in arteriolar diameter, represented
here as circles at the top of the graph, are primarily responsible for
the regulation of cerebral blood flow. The arterialarteriolar bed
actively constricts with increasing CPP and dilates with decreasing
CPP. Hypoperfusion occurs when CPP falls below the lower limit of
regulation. As CPP falls below this limit, the vascompression of the
arterioles and resistance increases markedly. The lower limit of
regulation is not precisely known and varies with age, with 4050
mm Hg for young pediatric cases (90) and 6070 mm Hg for an adult
(61,86).
CPP cause constriction of the arterial-arteriolar vascular

bed and lowering of ICP by a reduction in cerebral blood
volume. In addition, the resulting decrease of pre- and postcapillary pressure lessens fluid filtration and increases
absorption, thus reducing the effects of edema. The application of CPP-oriented therapy when autoregulation has
been lost may result in an imbalance of Starling forces at
the capillaries, leading to increased net filtration and
further brain injury by increased production of vasogenic
edema.
In contrast to clinical cardiology, where the physiological mechanism underlying the dynamic features of the
arterial pressure recording are fairly well understood, very
little is known about the mechanism that underlies the
shape of pulsations of intracranial pressure and how they
are influenced by changes in CPP. Past studies have indicated that pulsations in the cerebrospinal fluid develop
from either pulsations of the choroids plexus (92,93) or the
arterial vasculature (94,95). Depending on the dilatory
state of the cerebrovascular bed, ABP, ICP, and whether
the autoregulation is intact, at times the primary component of the ICP pulse may be due to pulsations of venous or
arterial origin (9496). Although the origin of the ICP
pulsation is not completely understood, possibly useful
methodologies involving the pulsation have been developed. Over two decades ago, Avezatt and van Eijnhoven
developed a procedure for distinguishing the occurrence of
the loss of regulation of cerebral blood flow through numerical analysis of the pulsation of intracranial pressure
(96,97) From laboratory studies, they found that within
the autoregulatory range, the relationship between the
mean amplitude of the pulsation of ICP increased linearly
with mean ICP with a high slope value. At mean ICP above
30 mm Hg, the slope of this relationship decreased. They
583
proposed that the flattening of the slope of this relationship

was an indication of loss of autoregulation (97). A weakness
of this technique is that it is dependent on heart rate (98).
A low heart rate reflects an increase in the volume, of
pulsatile cerebral blood inflow resulting from increased
cardiac stroke volume, and the converse is likely for high
heart rates. Thus, the variability in the amplitude of the
ICP due to heart-rate variability can complicate the analysis (98). However, one modification of the analysis technique is to examine the correlation value between
amplitude-pressure and mean ICP rather than the slope
of the respective regression line (99). Positive values of this
correlation value, which has been called RAP, indicate the
ability of the vasculature to regulate cerebral blood flow,
and negative values indicate poor cerebrovascular reserve
and impaired cerebrovascular reactivity (99,100). Statistical analysis revealed that patients with a fatal outcome
were associated with a negative RAP value (99). Most
recently, the RAP index has been used to predict the
achievable reduction in intracranial pressure a patient
can obtain by the implementation of moderate hyperventilation (101). A modification of the mean amplitude of ICP
and mean ICP characteristic has been proposed as a means
of predicting which patients with idiopathic adult hydrocephalus will have a good outcome after shunting surgery
(102). In this application, the amplitude of B-wave of ICP is
used instead of the amplitude of the ICP pulsation (102).
In addition to relatively synchronous pulsations of ICP
associated with the cardiac cycle, the ICP recording
obtained during mechanical ventilation contains a lowfrequency component at the rate of ventilation. Generally
during mechanical ventilation, inhalation is produced by
positive pressure and expiration occurs during zero pressure. Changes in pulmonary volume produce changes in
intrathoracic pressure that mechanically modulate arterial and venous blood flow and produce a cyclic compression
of the craniospinal sac. Evidence for this mechanical modulation by intrathoracic pressure can be observed in the
spectra of an ICP recording obtained during mechanical
ventilation. Because the pulsation in ICP associated with
the cardiac cycle produced is quasi-periodic and the rate of
ventilation is relatively constant, the spectra of the ICP
pressure recording contains salient peaks at the cardiac
frequency and its higher harmonics. In addition, each of
the spectra associated with cardiac cycle has sidebands
with a deviation at the ventilation frequency (103). Such a
result is consistent with the premise that intrathoracic
pressure changes mechanical modulated arterial and
venous blood pressure and volume of the craniospinal
sac (103).
Over the last few decades, several clinical and laboratory studies of the correlative relationship between ABP
and ICP have been completed. Portnoy et al. reported that
during normal vascular tone and intact regulation of cerebral blood flow, the ICP and ABP recordings do not look
similar, but during maximum vasodilation of the arterial
arteriolar bed and impaired autoregulation induced by
severe hypercapnia, these pressure recordings look
remarkably similar (104106). To numerically quantify
the correlation between ABP and ICP, this group examined
changes in the coherence function. Specifically, they found
584
that the frequency domain coherence function approached

unity when the ICP and ABP recording became similar.
Generally, their observations and the observations of
others using a spectral analysis systems approach have
suggested that the more similar the spectral components of
the ICP recording are to those of the ABP recording, the
more likely cerebral autoregulation is impaired (106108).
The physical process describing the ABP as the input to the
craniospinal sac and the ICP as the corresponding output
has been termed cerebrovascular pressure transmission
(106). An observational clinical study determined that four
types of frequency descriptions of the transmission characteristic could be identified. Two types were associated
with high ICP and the other two with low ICP (47).
More recent studies using time-domain correlation analysis on the ICP and ABP pressure recordings have been
completed. As correlation analysis of two signals in the
time domain is analytically equivalent to coherence analysis in the frequency domain, it was not unexpected that
studies on the normocapnic/hypercapnic piglet model
found that as the ICP and ABP recordings became more
similar, the maximum value of the correlation function
approaches unity, the pial arterioles became more dilated,
and cerebral blood flow increased (109,110). Consistent with
clinical reports that indicate that unlike adult patients,
brain-injured pediatric patients often demonstrate cerebral hyperemia and increased ICP (82,109,111), correlation
analysis of pressure recordings obtained from pediatric
patients have been found to often approach unity, indicating
the occurrence of inappropriate vasodilation and cerebral
hyperemia (112). Most recently, Czosnyka et al. have
reported the clinical use of a pressure reactivity index
(PRx). This index is a moving correlation coefficient between
40 consecutive samples of values for intracranial and arterial pressures averaged for a period of 5 s (113,114). They
have concluded that when slow waves in the ABP and ICP
recordings are present, the proposed clinical index provides
a continuous index of cerebrovascular reactivity (114,115).
In particular, the PRx was designed to evaluate the integrity
of the cerebrovascular response and estimate cerebrovascular autoregulatory reserve reactivity (114,115). The
hypothesis is that because of the sluggish nature of the
cerebrovascular system, naturally occurring slow-varying
oscillations of ABP can be used to evaluate the autoregulatory reserve reactivity (114,115). Unlike the coherence and
correlation indices described above, this index cannot be
related to a linear system model. By employing averaging
over a 5 s interval, most of the frequency changes above
0.2 Hz in the ABP and ICP recordings are filtered out. In
addition, Nyquists sampling theorem dictates that the highest frequency that can be represented by a signal sampled
every 5 s is 0.1 Hz or six oscillations per minute. As a result,
aliasing occurs, and the dynamical system relationship
between ABP and ICP is not precisely defined by this index.
Nevertheless, the PRx has been found to be a very useful
tool. Clinical observations demonstrate that the PRx is high
both during the occurrence of plateau waves and during
refractory ICP hypertension (115). In addition, the PRx has
been used to guide proposed therapies (116) and while
variable seems to provide a reliable index of autoregulation
(117).
Most recently, the regressive relationship between

mean ICP and mean CPP has been used to assess whether
autoregulation of cerebral blood flow is intact. In these
studies, ABP is pharmalogically elevated and the regressive relationship between ICP and CPP during the test
period is obtained (118). If pressure regulation of cerebral
blood flow is intact, then increases of CPP will cause
vasoconstriction, a decrease of ICP, and the regressive
relationship should have a negative slope parameter. A
marked positive slope parameter of this regressive relationship is an indication of passive pressure regulation;
increases of CPP cause dilation and increased ICP (119).
ANALYSIS METHODS BASED ON MODELS OF ICP
Models of ICP dynamics are based on the modified Monroe
Kellie doctrine (see the Physiology section), which assumes
that the total volume of intracranial substance, tissue,
blood, and CSF, is constant. Initial mathematical models
described the relationship between the formation of CSF
and absorption of CSF in equilibrium. Specifically, in these
models, laminar flow of CSF is assumed during steadystate conditions (120123) and the parameters of the
mathematical model are presented as an electrical circuit
(Fig. 2).
Using this circuit and manipulating the volume of CSF
by bolus either by withdrawal, injection, or constant infusion, it is possible to estimate Ro, C, the volumepressure
response, and the PVI. The latter two parameters have
been discussed previously in the Physiology section. All
parameters have been used to guide the management of
hydrocephalous and traumatic brain injury.
Higher order mathematical models of intracranial
hydrodynamics that incorporate the arterial and venous
blood volume compartments into the analog electric circuit
model have been developed to simulate the pulsatility,
which is present in the ICP recording. To account for the
reciprocal relationship between cerebral blood volume and
volume of CSF, Agarwal et al. (124) constructed a model
that connected vascular compliance to intracranial compliance in a series configuration. This modeling effort was
further developed in detail by Ursinos proposed fourthorder analog circuit model of overall human intracranial
CSF outflow
resistance
CSF storage
CSF formation
R
C
ICP
compliance
Ps
Dural Sinus
pressure
Figure 2. Electric circuit analog model of CSF system. In this

model, current represents flow of CSF fluid, and voltage represents
pressure. The capacitance C, represents intracranial compliance,
and the resistance R represents resistance of CSF fluid flow into
the venous system.
hydrodynamics (125,126). A feature of this model is that

the arterialarteriolar vascular bed consists of vasomotor
regulating resistance and compliance elements. Dilatory
and constrictive responses are primarily simulated by a
corresponding decrease or increase of vascular resistance.
This initial model has been modified by synchronously
modulating the terminal venous bed resistance to account
for cyclic variation of ICP produced by positive pressure
ventilation (127). The modified model demonstrates that
both the depth of modulation and the cerebrovascular
venous terminal bed resistance seem to be progressively
reduced with increasing levels of vasodilation induced by
increasing the levels of partial pressure of arterial blood
carbon dioxide. Numerical modeling of cerebrovascular
pressure transmission, the relationship between ABP
and CPP, has been used to assess changes in the modes
of pressure transmission before and after injury. Specifically, a proposed third-order model of ICP dynamics (128)
was used to define the mathematical structure to construct
a numerical identification model of cerebrovascular pressure transmission from laboratory pressure recordings
(112). Consistent with active vasoconstriction before brain
injury, during intact regulation of cerebral blood flow, the
highest modal frequency of cerebrovascular pressure
transmission decreased with increasing CPP. Conversely,
consistent with passive vasodilation with loss of autoregulation induced by fluid percussion injury, the highest modal
frequency demonstrated a direct relationship with increasing CPP (112).
SUMMARY
Although the subject of intracranial pressure dates back to
over 200 ago, both the invasive monitoring procedure
required to measure ICP with its accompanying risk of
infection and the nonunique nature of the extravascular
pressure measurement have retarded the development of
knowledge in this area. As a result, the intent of this article
is to provide the reader with a broad perspective of ICP
monitoring by providing background material on (1) the
physiological and anatomical characteristics of the craniospinal, (2) a history of instrumentation techniques
including the most recent advances, (3) a brief review of
significant clinical ICP monitoring studies with an emphasis on severe head injury, (4) a review of analysis methods,
and (5) models of ICP dynamics related to analysis.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
BIBLIOGRAPHY
22.
1. Stern WE. Intracranial fluid dynamics. The relationship of
intracranial pressure to the Monro-Kellie doctrine and the
reliability of pressure assessment. J Royal College Surgeons
Edinburgh 1963;9:1836.
2. Langfitt TW. Increased intracranial pressure. Clin Neurosurg 1969;6:436471.
3. Monro A. Observations on the Structure and Function of the
Nervous System. Edinburgh: Creech and Johnston; 1783.
4. Kellie G. An account of the appearances observed in the
dissection of two of three individuals presumed to have
perished in the storm of the third and whose bodies were
discovered in the vicinity of Leith on the morning of the 4th,
23.
24.
25.
26.
585
November 1821, with some reflections on the pathology of the

brain. Trans Med Chir Soc Edinb 1824;1:84169.
Weed LH, McKibben PS. Pressure changes in cerebrospinal
fluid following intravenous injection of solutions of various
concentrations. Am J Physiol 1919;48:512530.
Weed LH. Some limitations of the MonroKellie hypothesis.
Arch Surg 1929;18:10491068.
Kocher T. Hirnerschutterung, Hirndruck und chirurgische
Eingriffe bein Hirnerkrankungen Nothnagel: Spezielle
Pathologie und Therapie: 1901.
Duret H. Etudes experimentales et cliniques sur les traumatismes cerebrau. Paris: 1978.
Cushing H. Some experimental and clinical observations
concerning states of increased intracranial tension. Am J
Med Sci 1902;124:375400.
Cushing H. The blood pressure reaction of acute cerebral
compression, illustrated by cases of intracranial haemorrhage. Am J Med Sci 1903;125:10171044.
Cushing H. Concerning a definite regulatory mechanism of
the vasomotor center which controls blood pressure during
cerebral compression. Johns Hopk Hosp Bull 1901;12:290292.
Jennet WB. Experimental brain compression. Arch Neurol
1961;4:599607.
Johnston IH, Rowan JO. Raised intracranial pressure and
cerebral blood flow: 3. Venous outflow tract pressures and
vascular resistances in experimental intracranial hypertension. J Neurol Neurosurg Psych 1974;37:392402.
Browder J, Meyers R. Observations on behaviour of the
systemic blood pressure, pulse and spinal fluid pressure
following craniocerebral injury. Am J Surg 1936;31:403427.
Smyth CE, Henderson WR. Observations on the cerebrospinal fluid pressure on simultaneous ventricular and lumbar
punctures. J Neurol Psychiat 1938;1:226237.
Evans JP, Espey FF, Kristoff FV, Kimball FD, Ryder HW.
Experimental and clinical observations on rising intracranial
pressure. Arch Surg 1951;63:107114.
Marmarou A, Shulman K, LaMorgese J. Compartmental
analysis of compliance and outflow resistance of the cerebrospinal fluid system. J Neurosurg 1975;43:523534.
Miller JD, Stanek AE, Langfitt TW. Concepts of cerebral
perfusion pressure and vascular compression during intracranial hypertension. In: Meyer JS, Schade JP, editors. Progress in Brain Research. vol 35: Cerebral Blood Flow.
Amsterdam: Elsevier; 1972. pp 411432.
Miller JD, Garibi J, Pickard JD. Induced changes of cerebrospinal fluid volume: Effects during continuous monitoring of ventricular fluid pressure. Arch Neurol 1973;28:
265269.
Miller JD, Pickard JD. Intracranial volume/pressure studies
in patients with head injury. Injury 1974;5:265268.
Miller JD, Leech PJ, Pickard JD. Volume pressure response
in various experimental and clinical conditions. In: Lundberg
N, Ponten U, Brock M, editors. Berlin: Springer; Intracranial
Pressure II; 1975. pp 9799.
Miller JD. Volume and pressure in the craniospinal axis. Clin
Neurosurg 1975;22:76105.
Hase U, Reulen HJ, Meinig G, Schrmann K. The influence of
the decompressive operation on the intracranial pressure and
the pressure volume relation in patients with severe head
injuries. Acta Neurochir 1978;45:113.
Lundberg N. Continuous recording and control of ventricular
fluid pressure in neurosurgical practice. Acta Psychiat Neurol Scand 1960;36:149.
Baylis WM, Hill L, Gulland GL. On intracranial pressure and
the cerebral circulation. J Physiol 1897;18:334362.
Rosner MJ, Becker DP. ICP monitoring: Complications and
associated factors. Clin Neurosurg 1996;23:494519.
586
27. Maas A, Dearden M, Teasdale GM, Braakman R, Cohodan F,

Iannotti F, Karimi A, Lapirre F, Murray G, Ohman J, Persson
L, Servadei F, Stocchetti N, Unterburg A. EBIC-Guidelines for
management of severe head injury in adults. Acta Neurochir
1997;139:286294.
28. Naryan RK. et al. Guidelines for the management of severe
head injury. J Neurotrauma 1997;13:639734.
29. Gabe IT. Cardiovascular fluid dynamics. Acta Physiol Scand
1972;19:306322.
30. Piper IR, Dearden NM, Miller JD. Methodology of spectral
analysis of the intracranial pressure waveform in a head
injury intensive care unit. In: Hoff JT, Betz AL, editors.
Intracranial Pressure VII. Berlin: Springer-Verlag; 1989.
pp 668671.
31. Allan MWB. Measurement of arterial pressure using cathetertransducer systems. Br J Anaesth 1988;60:413418.
32. Czosnyka M, Czosnyka Z, Pickard J. Laboratory testing of
the Spiegelberg brain pressure monitor: A technical report. J
Neurol Neurosurg Psych 1997;63:732735.
33. Ostrup RC, Luerssen TG, Marshall LF. Continuous monitoring of intracranial pressure with a miniturized fibreoptic
device. J Neurosurg 1987;67:206209.
34. Marmarou A, Tsuji O, Dunbar JG. Experimental evaluation
of a new solid state ICP monitor. In: Nagai H, Kemiya K,
Ishiii S, editors. Intracranial Pressure IX. New York:
Springer-Verlag; 1994. pp 1519.
35. Morgalla MH, Cuno M, Mettenleiter H, Will BE, Krasznai L,
Skalej M, Bitzer M, Grote EH. ICP monitoring with a reusable transducer: experimental and clinical evaluation of the
Gaeltec ICT/b pressure probe. Acta Neurochir (Wien) 1997;139(6):
569573.
36. Holzschuh M, Woertgen C, Metz C, Brawanski A. Clinical
evaluation of the InnerSpace fibreoptic intracranial pressure
monitoring device. Brain 1988;12:191198.
37. Betsch HM, Aschoff A. Measurement artifacts in Gaeltec
intracrnial pressure monitors due to radio waves from personal beeper systems. Anaesthesiol Intensivmed Notfallmed
Schmerzther 1992;27(1):5152.
38. Munch E, Weigel R, Schmiedek P, Schurer L. The Camino
intracranial pressure device in clinical practice: Reliability,
handling characteristics and complications. Acta Neurochir
1998;140:11131119.
39. Chambers IR, Kane PJ, Choksey MS, Mendelow AD. An
evaluation of the Camino ventricular bolt system in clinical
practice. Neurosurgery 1993;33:866868.
40. Weinstable C, Richling B, Plainer B, Czech T, Spiss CK.
Comparative analysis between epidural (Gaeltec) and subdural (Camino) intracranial pressure probes. J Clin Monit
1992;8:116120.
41. Piper IR, Miller JD. The evaluation of the waveform analysis
capability of a new strain-gauge intracranial pressure microsensor. Neurosurgery 1995;36:11421145.
42. Signorini DF, Shad A, Piper IR, Statham PFX. A clinical
evaluation of the Codman microsensor for intracranial
pressure monitoring. Br J Neurosurgery 1988;12:223227.
43. Fernades HM, Bingham K, Chambers IR, Mendelow AD.
Clinical evaluation of the Codman microsensor intracranial
pressure monitoring system. Acta Neurochir Suppl (Wien)
1998;71:4466.
44. Mendelow AD, Rowan JO, Murray L, Kerr AE. A clinical
comparison of subdural screw pressure measurements with
ventricular pressure. J Neurosurg 1983;58:4550.
45. Piper IR, Barnes A, Smith DH, Dunn L. The Camino
intracranial pressure sensor: Is it optimal technology?
An internal audit with a review of current intracranial
pressure monitoring technologies. Neurosurgery 2001;49:
11581165.
46. Citerio G, Piper I, Cormio M, Galli D, Cazzaniga S, Enblad P,

Nilsson P, Contant C, Chambers I, on behalf of the BrainIT
Group, Bench test assessment of the new Raumedic.
47. Piper IR, Miller JD, Dearden NM, Leggate JR, Robertson I.
Systems analysis of cerebrovascular pressure transmission:
An observational study in head injured patients. J Neurosurg
1993;73:871880.
48. Becker DP, et al. The outcome from severe head injury with
early diagnosis and intensive management. J Neurosurg
1977;47:491502.
49. Marshall LF, Smith RW, Shapiro HM. The outcome with
aggressive treatment in severe head injuries. Part 1: The
significance of intracranial pressure monitoring. J Neurosurg
1979;50:2025.
50. Miller JD, et al. Significance of intracranial hypertension in
severe head injury. J Neurosurg 1977;47:503516.
51. Pitts LH, Kaktis JV, Juster R, Heilbrun D. ICP and outcome in
patients with severe head injury. In: Shulman K, Marmarou A,
Miller JD, Becker DP, Hochwald GM, Brock M, editors.
Intracranial Pressure IV. Berlin: Springer; 1980. pp 59.
52. Stuart G, Merry G, Smith JA, Yelland JDM. Severe head
injury managed without intracranial pressure monitoring. J
Neurosurg 1983;59:601605.
53. Miller JD, et al. Further experience in the management of
severe head injury. J Neurosurg 1981;54:289299.
54. Alberico AM, Ward JD, Choi SC, Marmarou A, Young H.
Outcome after severe head injury: Relationship to mass
lesions, diffuse injury and ICP course in pediatric and adult
patients. J Neurosurg 1987;67:648656.
55. Choi SC, Muizelaar JP, Barnes TY, Marmarou A, Brooks DM,
Young HF. Prediction tree for severely head injured patients.
J Neurosurg 1991;75:251255.
56. Vollmer DG, Torner JC, Jane J, et al. Age and outcome
following traumatic coma: Why do older patients fare worse?
J Neurosurg 1991;75:s37s49.
57. OSullivan MG, Statham PF, Jones PA, et al. Role of intracranial pressure monitoring in severely headinjured
patients without signs of intracranial hyperension on initial
computerised tomography. J Neurosurg 1994;80:4650.
58. Narayan RK, Greenberg RP, Miller JD, Enas GG, Choi SC,
Kishore PRS.
59. Saul TG, Ducker TB. Effect of intracranial pressure monitoring and aggressive treatment on mortality in severe head
injury. J Neurosurg 1982;56:498503.
60. Marmarou A, et al. Impact of ICP instability on outcome in
patients with severe head trauma. J Neurosurg 1991;75:s59
s66.
61. Jones PA, Andrews PJD, Midgley S, Anderson SI, Piper IR,
Tocher JL, Housley AM, Corrie JA, Slattery J, Dearden NM,
Miller JD. Measuring the burden of secondary insults in
headinjured patients during intensive care. J Neurosurg
Anaesth. 6:414.
62. Lewelt W, Jenkins LW, Miller JD. Autoregulation of cerebvral blood flow after experimental fluid percussion injury of
the brain. J Neurosurg 1980;53:500511.
63. Povlishock JT, Kontos HA. Continuing axonal and vascular
change following experimental brain trauma. Central Nervous System Trauma 1985;2:285298.
64. Nordstrom CH, et al. Cerebral blood flow, vasoreactivity and
oxygen consumption during barbiturate therapy in severe
traumatic brain lesions. J Neurosurg 1988;68:424431.
65. Miller JD, Adams JH. The pathophysiology of raised intracranial pressure. In: Adams JH, Duchen LW, editors. Greenfields Neuropathology. 5th ed. London: Arnold; 1992: 69105.
66. DeSalles AF, Muizelaar JP, Young HF. Hyperglycemia, cerebrospinal fluid lactic acidosis and cerebral blood flow in severly
headinjured patients. Neurosurgery 1987;21:4550.

67. Jaggi LJ, Obrist WD, Gennarelli TA, Langfitt TW. Relationship of early cerebral blood flow and metabolism to outcome in
acute head injury. J Neurosurg 1990;72:176182.
68. Muizelaar JP, Ward JD, Marmarou A, Newton PG, Wachi A.
Cerebral blood flow and metabolism in severely headinjured
children. Part 2: Autoregulation. J Neurosurg 1989;71:7276.
69. Muizelaar JP, Marmarou A, DeSalles AAF, Ward JD, Zimmerman RS, Zhongchao L, Choi SC, Young HF. Cerebral
blood flow and metabolism in severely headinjured children.
Part 1: Relationship with GCS score, outcome, ICP and PVI. J
Neurosurg 1989;71:6371.
70. Uzzell BP, Obrist WD, Dolinskas CA, Langfitt TW. Relationship of acute CBF and ICP findings to neuropsychological
outcome in severe head injury. J Neurosurg 1986;65:630635.
71. Rose J, Valtonen S, Jennett B. Avoidable factors contributing
to death after head injury. Brit Med J 1977;2:615618.
72. Gentleman D, Jennett B. Hazards of interhospital transfer of
comatose head injured patients. Lancet 1981;ii:835855.
73. Graham DI, Hume Adams J, Doyle D. Ischemic brain damage
in fatal nonmissile head injuries. J Neurol Sci 1978;39:213
234.
74. Graham DI, Ford I, Adams JH, Doyle D, Teasdale GM,
Lawrence AE, Mclellan DR. Ischemic brain damage is still
common fatal nonmissile head injury. 1989.
75. Miller JD, Garibi J. Intracranial volume/pressure relationships during continuous monitoring of ventricular fluid pressure. In: Brock M, Dietz H, editors. Intracranial Pressure.
Berlin: Springer; 1972. 270274.
76. Johnston IH, Rowan JO. Raised intracranial pressure and
cerebral blood flow: 3. Venous outflow tract pressures and
vascular resistances in experimental intracranial hypertension. J Neurol Neurosurg Psychi 1974;37:392402.
77. Yada K, Nakagawa Y, Tsuru M. Circulatory disturbance of
the venous system during experimental intracranial hypertension. J Neurosurg 1973;39:723729.
78. Nakagawa Y, Tsura M, Yada K. Site and mechanism for
compression of the venous system during experimental intracranial hypertension. J Neurosurg 1974;41:427434.
79. Harper AM. Autoregulation of cerebral blood flow: influence
of the arterial pressure on blood flow through the cerebral
cortex. J Neurol Neurosurg Psych 1966;29:398403.
80. Muizelaar JP, Becker DP. Induced hypertension for the
treatment of cerebral ischemia after subarachnoid hemorrhage. Direct effect on cerebral blood flow. Surg Neurol
1986;25:317325.
81. Obrist WD, et al. Cerebral blood flow and metabolism in
comatose patients with acute head injury. Relationship to
intracranial hypertension. J Neurosurg 1984;61:241253.
82. Bouma GJ, Muizelaar JP. Relationship between cardiac output
and cerebral blood flow in patients with intact and with
impaired autoregulation. J Neurosurg 1990;73:368374.
83. Chan KH, Miller JD, Piper IR. Cerebral blood flow at constant cerebral perfusion pressure but changing arterial and
intracranial pressure: Relationship to autoregulation. J Neurosurg Anaesth 1992;4:188193.
84. Piper IR, Lawson A, Dearden NM, Miller JD. Computerised
data collection: a microcomputer data collection system in head
injury intensive care. Br J Intensive Care 1991;1:7378.
85. Chambers IR, Treadwell L, Mendelow AD. The cause and
incidence of secondary insults in severely headinjured
adults and children. 2000.
86. Signorini DF, Andrews PJD, Jones PAJ, Wardlaw JM, Miller
JD. Prediciting survival using simple clinical variables: A case
study in traumatic brain injury. J Neurol Neurosurg Psych
1999;66:2025.
87. Signorini DF, Andrews PJD, Jones PAJ, Wardlaw JM, Miller
JD. Adding insult to injury: The prognostic value of early
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
99.
100.
101.
102.
103.
104.
105.
106.
587
secondary insults for survival after traumatic brain injury. J

Neurol Neurosurg Psychi 1999;66:2631.
Robertson CS, Valadka AB, Hannay HJ, Contant CF, Gopinath SP, Cormio M, Uzura M, Grossman RG. Prevention of
secondary ischemic insults after severe head injury. Crit
Care Med 1999;27(10):20862095.
Chambers IR, Jones PA, Lo TY, Forsyth RJ, Fulton B,
Andrews PJ, Mendelow AD, Minns RA. critical threshold
of intracranial pressure and cerebral perfusion pressure
related to age in paediatric head injury. 2005.
Piper I, et al. The BrainIT Group: Concept and core dataset
definition. Acta Neurochir 2003;145:615629.
Rosner MJ, Rosner SD, Johnson AH. Cerebral perfusion
pressure: management protocol and clinical results. J Neurosurg 1995;83:949962.
Bearing EA. Choroid plexus and arterial pulsation of cerebrospinal fluid. Arch Neurol Psych 1955;73:165172.
Hamit HF, Beall AC, DeBakey ME. Hemodynamic influences
upon brain and cerebrospinal fluid pulsations and pressures.
J Trauma 1965;5:174184.
Dardenne G, Dereymaeker A, Lacheron JM. cerebrospinal
fluid pressure and pulsatility. An experimental study of
circulatory and respiratory influences in normal and hydrocephalic dogs. 1969.
Hamer J, Alberti E, Hoyer S, Wiedemann K. Influence of
systemic and cerebral vascular factors on the cerebrospinal
fluid pulse waves. 1977.
Avezatt CJJ, van Eijndhoven JHM. Clinical observations on
the relationship between cerebrospinal fluid pulse pressure
and intracranial pressure. In: Cerebrospinal Fluid Pulse
Pressure and Craniospinal Dynamics: A Theoretical, Clinical and Experimental Study. Thesis. Erasmus Univ., Rotterdam, 1984.
Avezatt CJJ, van Eijndhoven JHM. Cerebrospinal fluid pulse
pressure and intracranial volumepressure relationships. J
Neurol Neurosurg Psych 1979;42:687700.
Daley ML, Gallo A, Mauch W. Analysis of the intracranial
pressure pulsation associated with the cardiac cycle. Innovation
Et Technology En Biologie Et Medicine 1986;7:537544.
Czosynka M, Guazzo E, Whitehouse M, Smielewski P, Czosynka Z, Kirkpatric P, Piechnik S, Pickard JD. Significance of
intracranial pressure waveform analysis after headinjury.
Acta Neurochir (Wien) 1996;138(5):531541.
Balestreri M, Czosynka M, Steiner LA, Schmidt E, Smielewski P, Matta B, Pickard JD. Intracranial hypertension:
What additional information can be derived from ICP waveform after head injury? Acta Neurochir 2004;146:131141.
Steiner IA, Balestreri M, Johnston AJ, Coles JP, Smielewski
P, Pickard JD, Menon DK, Czosnyka M. Predicting the
response of intracranial pressure to moderate hyperventilation Acta Neurochirurgica (online) 2005.
Lenfeldt N, Anderson N, AgrenWilsson A, Bergenheim AT,
Koskinen LO, Eklund A, Malm J. Cerebrospinal fluid pulse
pressure method: A possible substitute for the examination of
B waves. J Neurosurg 2004;101(6):944950.
Daley ML, Pasley R, Connolly M, Angel J, Timmons S, Stidham
G, Leffler C. Spectral characteristics of Bwaves and other low
frequency activity. Acta Neurochurirgica 2002;81:147150.
Chopp M, Portnoy H. System analysis of intracranial pressure. J Neurosurg 1980;53:516527.
Portnoy HD, Chopp M. Cerebrospinal fluid pulse wave form
analysis during hypercapnia and hypoxia. Neurosurgery
1981;9:1427.
Portnoy HD, Chopp M, Branch C, Shannon MD. Cerebrospinal fluid pulse waveform as an indicator of cerebral autoregulation. J Neurosurg 1982;56:666678.
588
107. Piper IR, Chan KH, Whittle IR, Miller JD. An experimental
study of cerebrovascular resistance, pressure transmission,
and craniospinal compliance. Neurosurgery 1993;32:805816.
108. Nichols JS, Beel JA, Munro LG. Detection of impaired
cerebral autoregulation using spectral analysis of intracranial pressure waves. J Neurotrauma 1996;13:439456.
109. Bruce DA, et al. Diffuse cerebral swelling following head
injury in children: The syndrome of malignant brain
edema. J Neurosurg 1981;54:170178.
110. Daley ML, Pasupathy H, Griffith M, Robertson JT, Leffler
C. Evaluation of autoregulation of cerebral blood flow by
correlation of arterial and intracranial pressure signals.
IEEE Trans Biomed Eng 1995;42:420424.
111. Bruce DA, et al. Pathophysiology, treatment and outcome
following severe head injury in children. Childs Brain
1979;5:174191.
112. Daley ML, Patterson S, Marmarou A, Leffler CW, Stidham
G. Pediatric traumatic brain injury: Correlation of intracranial and arterial pressure signals, Proc. 18th Annu. Int.
Conf. IEEE Engineering in Medicine and Biology Society,
Amsterdam, Netherlands, Nov. 1996.
113. Czosnyka M, Smielewski P, Kirkpatrick P, Laing R, Menon D,
Pickard J. Continuous assessment of the cerebral vasomotor
reactivity in head injury. Neurosurgery 1997;41:1119.
114. Czosnyka M, Smielewski P, Kirkpatrick P, Laing R, Menon
D, Pickard J. Continuous assessment of the cerebral vasomotor reactivity in head injury. Acta Neurochirugica
1998;71:7477.
115. Czosnyka M, Smielewski P, Piechnik S, Schmidt E, AlRawi
PG, Kirkpatrick PJ, Pickard JD. Hemodynamic characterization of intracranial pressure plateau waves in head
injured patients. J Neurosurg 1999;92:1119.
116. Steiner LA, Czosynka M, Piechnik SK, Smielewski P,
Chatfield D, Menon DK, Pickard JD. Continuous monitoring
of cerebrovascular pressure reactivity allows determination
of optimal cerebral perfusion pressure in patients with traumatic brain injury. Critical Care Med 2002;30: 733738.
117. Steiner LA, Coles JP, Johnston AJ, Chatfield DA,
Smielewske P, Fryer TD, Aigvirhio FI, Clark JC, Pickard
JD, Menon DK, Czosynka M. Assessment of cerebrovascular autoregulation in headinjured patients. Stroke
2003;34: 24042409.
118. Oertel M, Kelly DF, Lee JH, Glenn TC, Vespa M, Martin
NA. Is CPP therapy beneficial for all patients with high ICP.
Acta Neurochir 2002;81:6768.
119. Howells T, Elf K, Jones PA, RonneEngstrom E, Piper I,
Nilsson P, Andrews P, Enbald P. Pressure reactivity as a guide
in the treatment of cerebral perfusion pressure in patients with
brain trauma. J Neurosurg 2005;102(2):311317.
120. Marmarou A. A theoretical and experimental evaluation of
the cerebrospinal flid system. Ph.D. Dissertation, Drexel
University, 1973.
121. Marmarou A, Shulman K, LaMorgese J. Compartmental
analysis of compliance and outflow resistance of the cerebrospinal fluid system. J Neurosurg 1975;43:523534.
122. Marmarou A, Shulman K, Rosende RM. A nonlinear analysis of the cerebrospinal fluid system and intracranial
pressure dynamics. J Neurosurg 1978;48(3):332344.
123. Ekstedt J. CSF hydrodynamic studies in man. 1. Method of
constant pressure CSF infusion. J Neurol Neurosurg Psych
1977;40(2):105119.
124. Agarwal GC, Berman BM, Stark L. A lumped parameter
model of the cerebrospinal fluid system. IEEE Trans Biomed
Eng 1969;16:4553.
125. Ursino M. A mathematical study of human intracranial
hydrodynamics. Part 1the cerebrospinal fluid pulse pressure. Ann Biomed Eng 1988;16:379401.
126. Ursino M. A mathematical study of human intracranial

hydrodynamics. Part 2Simulation of clinical tests. Ann
Biomed Eng 1988;16:403416.
127. Pasley RL, Leffler CW, Daley ML. Modeling modulation of
intracranial pressure by variatrion of cerebral venous resistance induced by ventilation. Ann Biomed Eng 2003;31:
12381245.
128. Czosnyka M, Piechnik S, Richards HK, Kirkpatrick P,
Smielewski P, Pickard JD. Contribution of mathematical
modeling to the interpretation of bedside tests of cerebrovascular autoregulation. J Neurol Neurosurg Psych
1997;63: 721731.
See also BIOTELEMETRY;
HYDROCEPHALUS, TOOLS FOR DIAGNOSIS AND
TREATMENT OF; NEONATAL MONITORING; NEUROLOGICAL MONITORS.
MONITORING, NEONATAL. See NEONATAL

MONITORING.

AHMAD ELSHARYDAH
HAIBO WANG
RANDALL C. CORK
Louisiana State University
Health Center
Department of Anesthesiology
Shreveport, Louisiana
INTRODUCTION
In the neonatal intensive care unit (NICU), monitoring is
an integral part of patient care. The primary goal of
monitoring is to ensure that early and appropriate intervention can be initiated before to the onset of complications. Monitoring is also a means by which the effect of
interventions and therapies may be recorded, evaluated,
and controlled. The NICU staff have to deal with a full
range of conditions that can arise in the preterm or critically ill neonate, including hypoxemia, hypoglycemia,
hypotension, acidosis, and other serious problems. This
has led to the evolution and development of several monitors and different sensor-based technologies for use in
NICU monitoring including umbilical vessel monitoring.
These sensors may provide more accurate and reliable
monitoring of neonatal physiological and biochemical
changes with a rapid response time. The umbilical vessels
may be directly accessed in the first few days of life. An
umbilical artery catheter (UAC) may be used for blood
pressure monitoring, blood sampling, and fluid or drug
infusion. An umbilical vein catheter (UVC) may be used
for central venous pressure monitoring, blood sampling,
and fluid or drug infusion. Different types of commercially
available umbilical catheters are used for these purposes.
These catheters differ in their length, size, number of ports,
and their material (such as silicone and polyurethane).
Blood pressure (BP) monitoring is an important part of
neonatal intensive care both for the acutely ill and the
convalescing neonate. The most accurate method of measuring BP is by direct intra-arterial recordings, which
usually use an umbilical catheter to access the umbilical

artery. As blood gas measurement methods and monitors
have progressed in adult critical medicine, most of the new
techniques and sensors have been used in the NICU by
using umbilical artery catheterization. This article
addresses the potential benefits of umbilical vessel catheters and associated monitoring devices. It sheds light on
the catheters and monitors available on the market and
explains the complications and the risks of these catheters.
Furthermore, this article looks at the direction of this
technology in the future, and it tries to stimulate development of new technology for use in the monitoring of critically ill newborn infants (13).
Umbilical vein
589
Umbilical arteries
Historical Aspects
In 1946, Louis K. Diamond, a pediatrician from Boston, and
F. H. Allen, Jr. developed a technique that allowed blood
transfusion to take place through the infants umbilical
cord vein. Regular transfusions were difficult because of
the small size of blood vessels in newborns, and there was a
further complication due to the use of steel needles and
rubber catheters. Diamond used plastic tubing on the
umbilical vein, which was larger than average and
remained open for several days after birth (4). By the
1960s, electronic monitors came into use and blood gases
began to be measured. By the 1970s, the use of umbilical
catheters and arterial pressure transducers was routine
(5). The first organized NICU opened its doors at Yale-New
Haven Hospital in 1960. The first successful use of extracorporeal membrane oxygenation (ECMO) was in 1975.
ECMO eventually reduced infant mortality from 80% to
25% for the critically ill infants with acute reversible
respiratory and cardiac failure unresponsive to conventional therapy (6).
Anatomical and Physiological Aspects
The umbilical cord is a cordlike structure about 56 cm long,
extending from the abdominal wall of the fetus to the
placenta. Its chief function is to carry nutrients and oxygen
(O2) from the placenta to the fetus and return waste
products and carbon dioxide (CO2) to the placenta from
the fetus. It consists of a continuation of the membrane
covering the fetus and encloses a mucoid jelly (Whartons
jelly) with one vein and two arteries (7). Examination of the
umbilical cord (after cut) normally reveals two umbilical
arteries (UA) and one umbilical vein (UV) (Fig. 1). At skin
level, the UV is usually in the 12 oclock position and has a
thinner wall and wider lumen than do the UAs (2). Before
birth, blood from the placenta, about 80% saturated with
O2, returns to the fetus by way of the UV. On approaching
the liver, most blood flows through the ductus venous
directly into the inferior vena cava (IVC), short-circuiting
the liver. A smaller amount enters the liver sinusoids and
mixes with blood from the portal circulation. After a short
course in the IVC, it mixes with deoxygenated blood returning from the lower limbs before it enters the right atrium.
The blood leaves the heart to the descending aorta, where it
flows toward the placenta by way of the two UAs (7). The O2
saturation in the umbilical arteries is approximately 58%.
Changes in the vascular system at birth are caused by
Figure 1. Umbilical cord after it was cut.
cessation of placental blood flow and the beginning of

respiration. These changes are summarized in closure of
the umbilical arteries, closure of the umbilical vein and
ductus venous, significant reduction in the pulmonary
vascular resistance and right ventricle and right atrium
pressures, and significant increase in the systemic vascular resistance, left ventricle, and left atrium pressures
(8). After birth, the blood volume of the neonate is about
300 mL, the cardiac output averages 500 mL/min, and the
arterial blood pressure is about 70/50 during the first day,
which increases slowly over the next several months.
Arterial blood pressure of the neonate has been best correlated with birth weight. Moreover, systolic and diastolic
pressures in the neonate are significantly correlated to
blood pressure levels in the mother (9).
Indications and Contra-Indications for Umbilical
Artery Catheterization
The primary indications for umbilical artery catheterization include frequent or continuous blood gas measurements, continuous monitoring of arterial blood pressure,
and resuscitation (umbilical venous line may be the first
choice) (10). Secondary indications include infusion of
maintenance glucose-electrolyte solutions or medications,
exchange transfusions, angiography, and a port for frequent blood sampling, especially in a very low-birth-weight
neonate. These catheters should stay in place only as long
as a primary indication exists, with the exception of the
very low-birth-weight neonate who may need it for vital
infusions and frequent blood sampling. Contraindications
include evidence of local vascular compromise in lower
limbs or buttock areas, peritonitis, necrotizing enterocolitis, omphalitis, gastroischisis, and omphalocele.
Indications and Contra-Indications for Umbilical
Vein Catheterization
The most frequent indications for umbilical venous catheterization include emergency medication administration,
exchange transfusion, and partial exchange transfusion
(10). This catheter is also used for frequent blood sampling
and central venous pressure monitoring. Contraindications for the this catheter include routine fluid infusion
590
everything needed for umbilical catheterization, including

drapes, towels, suture, umbilical tape, skin preparation
materials, needles, forceps, syringes, and other instruments.
Umbilical Vessel Catheterization Procedure
Figure 2. Commercially available umbilical catheters: (a) singlelumen catheter and (b) dual-lumen catheter.
(relative contraindication), omphalitis, omphalocele, gastroischisis, necrotizing enterocolitis, peritonitis, and extrophy of the bladder.
Description of Available Catheters (Design and Material)
Catheters must be made of nontoxic materials that are the
least injurious to the vascular intima and least likely to
cause thromboses and early atherosclerotic lesions (11).
Commercially available umbilical catheters are usually
latex-free, made of silicon or high-quality aliphatic polyurethane elastomer (Tecoflex). Silicone catheters are soft,
not-irritating, usually not reactive to body tissues and body
fluids, not supportive of bacterial growth, and less problematic with blood clotting. Tecoflex is an advanced medical
formulation of polyurethane. Its physical characteristics
are very close to silicone; however, it is slightly stiffer
during insertion, which makes it easier to insert and better
to conduct arterial pressure. Tecoflex is thermosensitive,
softens at body temperature (12), and significantly reduces
the trauma to the vascular intima. These catheters have a
rounded tip, which makes them less likely to perforate
through the vessel during insertion. They have depth
markings at every centimeter for more accurate placement and encased radiopaque stripes to confirm placement
by X ray after placement. Umbilical catheters are available with single-lumen, dual-lumen, and triple-lumen
catheters (13) (Fig. 2,a,b) in different sizes ranging from
2.6 to 8.0 Fr. (Table 1) Umbilical artery catheter tips are
designed in two different ways: end-hole and side-hole tips.
The side-hole catheters use a special electrode to measure
blood gases and biochemicals. A Cochrane review by Barrington (14) showed that end-hole catheters are associated
with a much decreased risk of aortic thrombosis compared
with side-hole catheters. Therefore, umbilical artery catheters designed with a side-hole should not be used routinely
for umbilical artery catheterization in the newborn.
Furthermore, manufactures have made sterile, readyto-go umbilical catheterization trays. These trays contain
Table 1. Neonate Weight and Umbilical Catheter Size

Neonate Weight (g)
<1500 g
>1500 g
UAC Size (Fr)
UVC Size (Fr)
3.5 Fr
5.0 Fr
3.5 Fr
5.0 Fr
The infant must be supine and restrained (2). A field

around the umbilicus is sterilized and draped, and a silk
suture is looped around the base of the umbilical stump.
The distal end of the stump is cut off, leaving 2 cm of stump,
and the vessels are occluded to prevent blood loss. For the
umbilical artery catheterization (15), the stump is firmly
grasped with the gloved fingers of one hand, and one of the
two thick-walled umbilical arteries is dilated with a curved
iris forceps; then the umbilical artery catheter is inserted
into the artery. Some resistance may be encountered when
the catheter has been advanced 3 to 5 cm into the vessel,
but this resistance can usually be overcome by applying
steady downward pressure on the catheter. If the catheter
cannot advance, a second catheter can be inserted into the
other artery while leaving the first catheter in place. This
maneuver often causes one or the other vessel to relax and
permits one catheter to be advanced into the aorta.
Advancement of an UAC should place the tip above the
celiac axis but below the ductus arteriosis. All air should be
removed from the system. The accidental injection of small
amounts of air (<0.1 mL) may obstruct blood flow to the
legs for several hours. The catheter should be attached to a
pressure transducer and the arterial pressure measured.
For the umbilical vein catheterization, the single, large,
thin-walled umbilical vein is grasped with an iris forceps,
and the air-free catheter, which is connected to a closed
stopcock, is inserted 3 to 5 cm into the vessel with a twisting motion. The UVC tip should lie a few centimeters into
the umbilical vein or inferior vena cava. The stopcock must
be closed to prevent aspiration of air through the catheter
should the patient take a deep breath. It is imperative that
no air be injected through venous catheter, because the air
may enter the systemic circulation through the foramen
ovale and occlude a coronary or cerebral artery. If it does,
the neonate may die or suffer central nervous system
damage. If the catheter tickles the atrial septum, the
neonate may suffer arrhythmias. Withdrawal of the catheter a
short distance can solve the problem. Plain radiographs
should be taken to confirm placement (Fig. 3). High placement of the UAC is defined in one major review of the literature
as one with the tip in the descending aorta above the level of
the diaphragm and below the left subclavian artery and low
placement of the UAC as one with the tip above the aortic
bifurcation and below the renal arteries (16).
Neonatal Blood Gas and Biochemical Measurement
The blood gas measurement is the most widely used clinical
method for assessing pulmonary function in the neonate. It
forms the basis for diagnosis and management of neonates
with cardiorespiratory disease (17). The physiology of blood
gases is discussed in other parts of this Encyclopedia. We
will discuss in this section some issues related to the
neonatal blood gas measurement. The dissociation curve
of fetal hemoglobin (as compared with adult) is shifted to
591
and electrolyte balance, and the neonates small size make

the electrolytes and glucose assessment difficult and complicated (20), especially in the first week of life. A physiologic decrease in extracellular water volume, as well as a
transient increase in serum potassium and transient
decreases in plasma glucose and total plasma ionized
calcium concentrations, must be taken into account when
monitoring neonatal electrolytes and glucose. Frequent
and even continuous monitoring of physiological parameters is indicated in some cases, including glucose and
calcium monitoring in the premature newborn (limited
hepatic glycogen storage) and in newborns of diabetic
mothers (21). Before birth, fetal glucose is slightly higher
than maternal glucose. With cord clamping, neonatal
plasma level plummets over the first 6090 min of life
(23), (23). Neonatal hormonal changes later leads to an
increase in endogenous glucose production and stabilization of its level.
Technical Aspects of Neonatal Blood Gas
and Biochemical Measurement
Figure 3. Anteroposterior roentgenogram shows the position of
umbilical artery and vein catheters. Lateral roentgenogram is
needed to distinguish the umbilical artery from the umbilical
vein catheter and to determine the appropriate level of
insertion. A endotracheal tube; B umbilical venous catheter.
C umbilical artery catheter passed up the aorta to T12.
the left, and at any arterial O2 partial pressure (PaO2)

below 100 mm Hg (13332.2 Pa) fetal blood binds more O2.
This shift seems to be the result of the lower affinity of fetal
hemoglobin for 2,3-diphosphoglycerate (DPG). Shunting is
a common occurrence in the neonate, such as in congenital
cyanotic heart disease, persistent fetal circulation, or
atelectasis. O2 supplementation does not prevent the
hypoxia produced by such a shunt. Arterial carbon dioxide
partial pressure (PaCO2) is an important measure of pulmonary function in neonatal respiratory disease. The
initiation of ventilation with the first breath after normal
delivery results in a rapid fall in PaCO2 within minutes of
birth. PO2 rises rapidly to levels of 60 to 90 mm Hg (17).
Immaturity of the kidney in the newborn affects the basal
acidbase status and the response to additional acid and
alkali loads (18). The blood bicarbonate (HCO3) concentration is typically lower than in the adult (1821 mEq/L).
However, the blood pH (7.35 7.43) is only marginally
decreased because of the compensatory increase in the
neonatal respiratory rate. Table 2 lists the normal values
of pH, PaCO2, and total CO2 in the adult and in preterm
and term neonates (19). The transition from fetal to neonatal life, which is associated with rapid changes in fluid
Table 2. Acidbase Parameters in Neonates and Adults
(mean SD)
pH
PCO2
Total CO2
Preterm
7.40 0.08
34.0 9.0
21.0 2.0
Term
7.40 0.06
33.5 3.6
21.0 1.8
Adult
7.40 0.03
39.0 2.6
25.2 2.8
Technologic innovations in the development of biosensors

and microprocessors have led to development of bedside
small and accurate point-of-care (POC) devices (24). POC
devices have been used widely in critical care units, including the NICU. These devices are divided into two groups:
Analyzers, which are not attached to the patient blood
source and require blood sampling, and monitors, which
are continuous or near-continuous patient-attached POC
monitors (25). Neonatal biochemical measurement may
include several important blood parameters, such as
sodium, potassium, calcium, glucose, and even lactate
Table 3.
Intermittent Blood Gas and Biochemical Measurement
(Sampling). This is the most common technique used in
the NICU for invasive neonatal blood gas and biochemical
measurement. Usually a small blood sample is withdrawn
from a blood vessel, such as an umbilical vessel through an
umbilical catheter. This sample is analyzed by using a
bedside point-of-care analyzer or sent to a small satellite
laboratory unit in the NICU or to the central laboratory.
Blood gas analyzers are discussed in other articles in this
Encyclopedia. These analyzers use the principles of Clarks
electrode for PO2 measurement and Severinghauss
electrode for PCO2 measurement. Some blood-sample collecting systems are commercially available, such as the
Edward VAMP Jr. system manufactured by Edward
(Fig. 4), which are manufactured specifically for neonate
and small children use. This system is latex-free, disposable, closed, with small volume systems. Such systems are
designed to the decrease the risks of blood loss, infection,
and air bubbles (26).
Continuous Intravascular Neonatal Blood Gas and
Biochemical Sensors. There are several drawbacks for
using frequent arterial blood gas (ABG) sampling in the
neonate (27). This method may result in blood loss that can
necessitate blood transfusion. Moreover, in this method,
rapid changes in blood gas values may be missed, especially
592
Table 3. The Main Blood-Chemistry Parameters Monitored in the Neonatal Care Unit, With Typical Sensing Principles and
Transducers
Parameter
Invasive Blood Pressure
PO2
PCO2
Glucose
Lactate
Electrolytes (K, Na, Ca, Cl)
pH
Hemoglobin
Sensing Principle(s)
Electrical, impedance
Optical, reflection
Optical, fluorescent
Electrochemical, amperometric
Optical, fluorescent
Electrochemical, potentiometric
Optical, colorimetric
Optical, absorption
in conditions needing quick and close ABG monitoring,

such as after surfactant administration (28) and during
high-frequency ventilation (29). These drawbacks dictate
the need for a more efficient real-time way to monitor ABGs
(30). For the last two decades, intra-arterial PaO2 monitoring has been available with the use of a Clark electrode (31)
Figure 4. Edward VAMP Jr. blood-sample collecting system.
Transducer(s)
Strain gauge, piezoresistor
Photodetector and emitter
Photomultipler tube
Clark oxygen electrode
Photomultipler tube
Ion-sensitive electrode
Photodetector
Enzyme modified biosensor
Photodetector
Enzyme modified biosensor
Photodetector
Ion-selective electrode (ISE)
Photodetector
Ion-sensitive electrode
Photodetector and emitters
or a multiparameter sensor with an umbilical artery catheter (32). New fiber-optic continuous blood gas monitoring
sensors have been validated and used in the neonate with
an UAC, such as Neotrend. These devices promise to be
safe, easy to use, and accurate in newborns. However, the
cost-effectiveness of these devices is still not well established (33) (refer to the blood gas measurement article in
this Encyclopedia for details about Neotrend). The ex vivo
in-line VIA Low Volume Mode blood gas and chemistry
monitoring system (VIA LVM Monitor; Metracor Technologies, Inc., San Diego, CA) is an in-line, low-volume POC
monitor for neonates and children. Studies have shown
promising results in using this monitor in the neonate.
However, its cost-effectiveness has not been established yet
(25,34). This device measures pH, PaCO2, PO2, Na, K,
and hematocrit (Hct) by automatically drawing blood
(almost 1.5 mL) from a patients arterial catheter, analyzing it, and reinfusing the blood sample back into the
patient. Results are usually displayed in 12 min.
The operator performs an initial calibration, and then
the device performs self-calibration after each sample
and at least every 30 min. This machine is compatible with
all sizes of UACs and peripheral arterial catheters. Figure
5 shows a diagram of the VIA LVM monitor and its components at the neonate bedside.
Figure 5. Diagram of the VIA LVM in-line ex vivo monitor and

its components at the neonate bedside.
Neonatal Hemodynamic Monitoring

Direct arterial blood pressure monitoring is the most accurate
technique for determining arterial pressure in the neonate
(9). This method is best done by using umbilical artery
catheterization. It is an easy and quick procedure in comparison with other neonatal artery catheterizations, such as
radial and femoral artery catheterization. After the umbilical artery catheterization is done as described above, it is
connected to a pressure transducer and a continuous flow
device, as well as stopcock and manometer tubing.
Basic Concepts. Hemodynamic pressure monitoring
requires several basic components to accurately measure
the physiologic pressures. These components are: (1) an
intravascular catheter, (2) connecting tubing and stopcocks to connect that catheter and the patients blood
vessels to the monitoring system, (3) a pressure transducer
to convert the mechanical impulse of a pressure wave into
an electrical signal through movement of a displaceable
sensing diaphragm, (4) a continuous flush device that fills
the pressure tubing with fluid and helps prevent blood from
clotting in the catheter, (5) an amplifier that increases the
low-voltage signal from the pressure transducer to a signal
that can be displayed on a display device, (6) an oscilloscope
to display waveforms and a digital readout to display
numerical data, and (7) a processor or microcomputer that
is used to calculate various hemodynamic parameters
based on the measured variables.
Pressure Transducers. Pressure transducers are divided
in two groups: (1) External transducers located away from
the intravascular catheter and connected to that catheter
via fluid-filled pressure tubing, and (2) catheter-tip transducers. The external transducers use three types of sensing
elements: (1) strain gauges. These consist of an electrically
conductive elastic material that responds reversibly to
deformation by a change in electrical resistance. The resistance is converted into a voltage signal by connecting the
elements to form a Wheatstone bridge circuit. The output
voltage is proportional to the applied pressure and the
excitation voltage. Strain gauges are the most common
method of pressure transduction. (2) Silicon strain gauges.
These are thin slices of silicon crystal bonded onto the back
of a diaphragm. The movement of the diaphragm causes a
change in the resistance of the crystal, which can be
converted into an output signal. Silicon strain gauges
are more sensitive than standard strain gauges, but they
are affected by temperature and are non-linear. (3) Optical
sensors: These are also diaphragms, but in this case, the
movement of the diaphragm is sensed by reflecting a beam
of light off the silver back of the diaphragm onto a photoelectric cell. The intensity of light sensed by the photoelectric cell changes with the diaphragm position, causing
a decrease in its electrical output.
The Pressure Measurement System. The arterial waveform can be characterized as a complex sine wave, which is
the summation of a series of simple sine waves of different
amplitude and frequencies. The fundamental frequency (or
first harmonic) is equal to the heart rate. The first 10
593
harmonics of the fundamental frequency contribute to

the waveform. Any measurement system responds to a
restricted range of frequencies only. Within this range,
the system may respond more sensitively to some frequencies than to others. The response of the system plotted
against the signal frequency is the frequency response of
the system. However, the measurement system may possess natural frequencies or resonances determined by the
inertial and compliant elements in a mechanical system.
These resonances can distort the output signals. Therefore,
it is essential that the natural frequencies do not lie in the
operating frequency range of the instrument. Moreover,
the output signal of a measurement may differ from the
input signal created by the bloodstream because of the
inertial components, frictional effects of movement, viscous
forces of fluids, and electrical resistance. The property that
determines these effects is called the damping of the
system (35).
Technical Management of Pressure Monitoring System.
These are some significant practical points in operating
this system:
1. Removal of all air bubbles from system: Air is more
compressible than fluid, and it tends to act as a
shock absorber within the pressure monitoring
system, leading to an overdamped waveform, which
may lead to false readings of the blood pressure.
Moreover, air bubbles may cause serious air embolism, especially in neonates and small children.
2. Zeroing the transducer: The accuracy of invasive
pressure measurements is dependent on the establishment of an accurate reference point (Zeroing).
This is done by opening the stopcock to atmospheric
pressure and zeroing the measurement system to
eliminate the effect of the atmospheric pressure,
and by leveling the transducer to the level of the
upper portion of the right atrium (the patients midaxillary line or phlebostatic axis) to eliminate the
effect of the blood hydrostatic pressure.
3. Fast-flush technique: A fast-flush or square wave
test is performed by opening the valve of the continuous flush device, which leads to an acute increase
in the fluid flow rate through the catheter-tubing
system from the usual 13 mL/h to 30 mL/h. This
generates an acute rise in pressure within the system
such that a square wave is generated on the monitor.
With closure of the valve, a sinusoidal pressure wave
of a given frequency and progressively decreasing
amplitude is generated. A system with appropriate
dynamic response characteristics will return to the
baseline pressure waveform within one to two oscillations. If the fast-flush technique produces dynamic
response characteristics that are inadequate, the
clinician should troubleshoot the system (i.e., remove
all air bubbles, minimize tubing length and stopcocks, etc.) until an acceptable dynamic response is
achieved. The above-explained basic concepts in
hemodynamic monitoring are used in pressure monitoring in neonates as well as in adults. Because of
594
Figure 6. Neonatal/Pediatric Deltran pressure tubing with

needleless blood collection system.
the small size of a neonate (or premature newborn)

and because of the different indications and expected
complications from adults and older children, these
monitors have been modified to fit these requirements. These modifications and changes include (1)
using a more simple tubing system with fewer stopcocks to reduce the risk of air embolism and infection,
(2) using a smaller volume tubing system with small
syringes to decrease blood loss and the need for blood
transfusion, and (3) using a special constant-low-rate
flush system to decrease the risk of fluid overloading.
After correct placement of the umbilical artery catheter, a stopcock (free of air bubbles) is connected to its
distal end. Then a fluid-filled, well-flushed pediatric/
neonatal blood pressure tubing is connected to that
stopcock (or directly to the catheter). The transducer
is zeroed and leveled to the midaxillary level. There
are several commercially available neonatal/pediatric pressure transducers and tubing systems. Figure 6
shows a disposable, latex-free neonatal/pediatric
Deltran blood pressure monitoring and needleless
blood collection system (36). For central venous pressure (CVP) monitoring, a dual-lumen or triple-lumen
umbilical vein catheter may be used. Figs. 2 and 7
show some of the currently available catheters. Neonatal/pediatric pulmonary artery catheters (PACs)
have been used through the umbilical vein. However,
they are difficult to place and have numerous complications and risks associated with their placement.
Other noninvasive cardiac monitors, such as echocardiography, can assess cardiac output and other
physiological cardiac parameters. Thus, use of the
neonatal PAC has decreased significantly. The most
common indications for pediatric PACs are for cardiogenic shock, for severe distributive shock, for the
use of very high ventilator pressures to achieve ade-
Figure 7. Catheter.
quate oxygenation, and for the perioperative management of patients who have undergone complex
cardiac or other major surgeries (37). The smallest
thermodilution catheter available is 5 Fr in size,
although a single-lumen 4-Fr catheter exists and is
useful for measuring pulmonary artery pressure (PAP)
or obtaining mixed venous saturation (MvO2). Potential complications of pulmonary artery catheterization
include pulmonary artery erosion or infarction, dysrhythmia, damage to the pulmonic valve, coiling in the
right ventricle, and cardiac perforation (38).
Complications and Risks
Umbilical Artery Catheter. Many complications and
risks are associated with UAC placement (39). Therefore,
these catheters should not be used solely for fluid and
medication administration. If an infant does not require
frequent arterial blood sampling or continuous blood pressure monitoring, there is almost no justification for leaving
a UAC inserted. The advantages and disadvantages of
high versus low UAC are still debated. Recent studies
(4043) have found that high catheters are associated
with a decreased incidence of complications without a
statistically significant increase in any adverse sequelae.
Therefore, one major review of the literature has concluded
that there appears to be no evidence to support the use of
low placed umbilical artery catheters. High catheters
should be used exclusively (40). It is clear that UACs that
are located between the high and low positions are never
appropriate. Catheters in these positions have been associated with refractory hypoglycemia (infusion into the celiac
axis), paraplegia (infusion into the artery of Adamkievicz)
595
Table 4. Care of Indwelling Umbilical Catheter

Change all tubings and connections daily.

Secure and label all tubing, and connections should be secured and labeled appropriately.
Use only appropriate filters.
Maintain catheter, connections, and tubing free of blood to prevent clot formation, or inadvertent flushing of preexisting clots into the
neonate.
Flush catheter with 0.5 mL of flush solution each time blood sample is drawn.
Chart fluids infused in intake/output record.
Infuse heparinized parenteral solution continuously through catheter, interrupting only to obtain blood samples.
(44), and thromboses that affect the kidneys (infusion into

the renal arteries) or the gut (infusion into the mesenteric
arteries). A catheter that is found in this intermediate
position should be pulled to a low position or removed.
Similarly, catheters should not be placed below the level of
L5 because of the risk of gluteal skin necrosis (45) and sciatic
nerve damage (45). Catheters that are placed below the level
of L5 should be removed promptly. Other complications may
include catheter occlusion, infection, air embolism, breaks
or transection of the catheter, electrical hazards, intravascular knots in the catheters, bladder injury, peritoneal
perforation, Whartons jelly embolus, and others.
Umbilical Venous Catheter. Thromboembolic events
and infections are common complications of UVC use.
These complications are similar to those of other central
catheters, although UVCs are associated with an increased
risk of localized infections of the liver and heart. Complications that are specific to UVC commonly are the result of
malposition of the catheter. Nearly all experts recommend
placement of the catheter outside the heart in the IVC (46).
Complications occur as a result of placement of the catheter
in the right side of the heart or the left side (via the foramen
ovale). Cardiac arrhythmias are common complications,
but these arrhythmias usually resolve after catheter withdrawal from the heart. Cardiac perforation with subsequent pericardial effusion and cardiac tamponade has been
reported (47) Quick diagnosis (high index of suspicion,
chest radiograph, and ultrasonography) and prompt treatment with pericardiocentesis decrease mortality significantly in these neonates. Placement of the catheter in the
portal system can result in serious hepatic injury. Hepatic
necrosis can occur from thrombosis of the hepatic veins or
infusion of hypertonic or vasospastic solutions into the liver.
Necrotizing enterocolitis and perforation of the colon also
have been reported after positioning of the catheter in the
portal system. Other complications of umbilical venous
catheters have been reported and include perforation of
the peritoneum, electrical hazards, and digital ischemia.
Umbilical Catheter Removal and Maintenance
According to the American Academy of Pediatrics (AAP)
guidelines (48), umbilical artery or venous catheters
should be removed and not replaced if there is any sign
of catheter-related bloodstream infections, vascular insufficiency, or thrombosis. Moreover, umbilical catheters must
be removed as soon as possible when no longer needed or
when any sign of vascular insufficiency to the lower extremities is observed. An umbilical artery catheter should not
be left in place more than 5 days. However, an umbilical
venous catheter can be used up to 14 days if managed

aseptically. Umbilical venous catheters may replaced only
if the catheter malfunctions. Table 4 lists the important
tasks for daily care of umbilical indwelling catheters.
FUTURE TRENDS IN INVASIVE NEONATAL MONITORING
Providing real-time, accurate, reliable, compact, at the bedside, and safe neonatal monitoring devices is the future goal
of researchers and manufactures involved with neonatal
critical care (3). However, the cost issue is still outstanding
and is going to be a major factor in directing this industry.
The market demand for integrated critical care units and the
clinical demand for continuous and rapid biochemical monitoring at the bedside, especially in critically ill patients, will
lead to a closer integration between the vital signs monitors
and POC analyzers. The monitor and POC analyzer manufacturers have been separate companies. This will slowly
change through the formation of strategic alliances and
mergers. Therefore, new devices will combine continuous
vital signs, blood gases, and blood chemistry in the critical
care units, including the NICU. The demand for continuous
monitoring of biochemical parameters is at the same time
bounded by the requirement for low-blood-loss systems, especially in the NICU. This is where the in-line and indwelling
analyzers can offer a major advantage over the POC analyzers. Fortunately, technology is also moving in the right
direction to minimize iatrogenic blood loss and decrease the
risk of infection through the advances made in sampling,
preparation, and handling of liquids using microfluidic
techniques and closed systems (49). It is likely that in-line
monitors that interface to umbilical artery catheters will
become more widespread and will require decreasing sample-volumes. The continuing application and refinement of
established optical assay methods, such as absorption and
fluorescence spectroscopy, onto fibre-optic cables will enable
the detection of increasing numbers of analytes by indwelling probes encased within the umbilical catheters (50,51).
SUMMARY
The use of umbilical vessel catheterization and associated
monitoring techniques and devices has been advanced
dramatically since Dr. Louis K. Diamond of Boston used
plastic tubing on the umbilical vein for blood transfusion in
1946. Advances in invasive neonatal monitoring and neonatal intravascular access have led to a significant reduction in the incidence of complications and have increased the
number of indications for umbilical catheters. Umbilical
vessel catheterization is now a routine and safe procedure
596
in the NICU. Moreover, with the increase in the survival

rate of low- and very low-birth-weight neonates, the need for
these catheters has increased. Sometimes these catheters
are the only intravascular access that can be established in
this new group of patients. The innovation and development
of medical applications of silicone and polyurethane enable
the manufacturers to make soft, small catheters with adequate lumen size. These catheters cause minimal adverse
reaction to the newborn body. To decrease the risk of blood
loss and infection, new blood sampling devices have been
developed. These systems are closed and have small volume
tubing. New technologies for continuous monitoring of blood
gases and neonatal chemistries in-line have been integrated
with the umbilical catheters by using special sensors
encased in the tip of the catheters. The future trend is to
use smaller, compact, real-time, bedside monitors combined
with pulse oximetry and other vital signs monitors. However, the issue of the cost-effectiveness of these machines
has yet to be determined.
BIBLIOGRAPHY
1. Walsh-Sukys MC, Fanaroff AA. Perinatal services and
resources. In: Fanaroff AA, Martin RJ, editors. NeonatalPerinatal Medicine, Diseases of the Fetus and Infant. 6th ed.
St. Louis: Mosby; 1999.
2. Stovroff M, Teague WG. Pediatric surgery for the primary
care pediatrician, Part II: Intravenous access in infants and
children. Pediatric Clin North Am 1998;45(6).
3. Murkovc I, Steinberg MD, Murkovc B. Sensors in neonatal
monitoring: Current practice and future trends. Technol
Health Care 2003;11:399412.
4. Diamond LK, Allen Jr FH, Thomas Jr WO, Erythroblastosis
fetalis. VII. Treatment with exchange transfusion. N Engl J
Med 1951;244:3949.
5. Klaus MH, Fanaroff AA. Care of the High-Risk Neonate, 5th
ed. Philadelphia: WB Saunders; 2001
6. Bartlett RH, Roloff DW, Cornell RG, Andrews AF, Dillon PW,
Zwischenberger JB. Extracorporeal circulation in neonatal
respiratory failure: A prospective randomized study. Pediatrics 1985;76:479497.
7. Sadler TW. Langmans Medical Embryology: Cardiovascular
System, 8th ed. Philadelphia: Lippincott Williams & Wilkins; 2000.
8. Guyton AC, Hall JE. Textbook of Medical Physiology: Fetal
and Neonatal Physiology, 10th ed. Philadelphia: W.B. Saunders; 2000.
9. Zahka KG. Principles of neonatal cardiovascular hemodynamics. In: Fanaroff AA, Martin RJ. editors, Neonatal-Perinatal Medicine, Diseases of the Fetus and Infant. 6th ed. St.
Louis: Mosby; 1999.
10. Grady M, Procedures. In: Gunn VL. Nechyba C, editors,
Harriet Lane Handbook: A Manual for Pediatric House Officers, 16th ed. St. Louis: Mosby; 2002.
11. Brown EG, Krouskop RW. Monitoring, umbilical artery and
vein. In: Webster JG, editor. Encyclopedia of Medical Devices
and Instrumentation, New York: Wiley; 1988
12. Chidi CC, King DR, Boles Jr ET, An ultrastructural study of
the intimal injury induced by an indwelling umbilical artery
catheter. J Pediatr Surg 1983;18:109115.
13. http://www.utahmed.com/umbili-c.htm.
14. Barrington KJ. Umbilical artery catheters: Catheter design
(CochraneReview).In: TheCochraneLibraryIssue4.Oxford;1997.
15. Gregory GA. Resuscitation of the newborn. In: Miller RD,
editor. Millers Anesthesia. 6th ed. New York: Elsevier; 2005.
16. Barrington KJ. Umbilical artery catheters in the newborn:

effects of position of the catheter tip. Cochrane Database Syst
Rev CD000505, 2000.
17. Carlo WA. Assessment of pulmonary function. In: Fanaroff
AA, Martin RJ, editors. Neonatal-Perinatal Medicine, Diseases of the Fetus and Infant. 6th ed. St Louis: Mosby; 1999.
18. Stork JE, Stork EK. Acid-base physiology and disorders in
the neonate. In: Fanaroff AA, Martin RJ. editors, NeonatalPerinatal Medicine, Diseases of the Fetus and Infant. 6th ed.
St. Louis: Mosby; 1999.
19. Lorenz JM, Kleiman LI, Kotagal UR, Reller MD. Water
balance in very low birth infants: Relationship to water and
sodium intake and effect on outcome J Pediatrics 1982;101:
423432.
20. Lorenz JM. Assessing fluid and electrolyte status in the
newborn, Nat Acad of Clin Biochem. Clin Chem
1997;43(1):205210.
21. Heck LI, Erenberg A. Serum glucose values during the first
48 h of life. J Pediatr 1978;110:119122.
22. Srinivasan G, Pildes RS, Cattamanchi G, Voora S, Lillian LD.
Plasma glucose values in normal neonates: A new look. J
Pediatr 1984;105:114119.
23. Tsang RC, Chen IW, Freidman MA, Chen I. Neonatal parathyroid function: role of gestational and postnatal age. J
Pediatr 1973;83:728730.
24. Ehrmeyer SS, Laessig RH, Leinweber JE, Oryall JJ. Medicare/CLIA final rules for proficiency testing: Minimum intralaboratory performance characteristics (CV and bias) needed
to pass. Clin Chem 1990;36:17361740.
25. Billman GF, et al. Clinical performance of an in-line, ex vivo
point-of-care monitor: A multicenter study. Clin Chem
2002;48(11): 20302043.
26. http://www.edwards.com/Products/PressureMonitoring/
Vamp Jr.htm
27. Meyers PA, Worwa C, Trusty R, Mammel MC. Clinical
validation of a continuous intravascular neonatal blood gas
sensor introduced through an umbilical artery catheter.
Respir Care 2002;47(6):682687.
28. Kresch MJ, Lin WH, Thrall RS. Surfactant replacement
therapy. Thorax 1996;51(11):11371154.
29. Nelle M, Zilow EP, Linderkamp O. Effects of high-frequency
oscillatory ventilation on circulation in neonates with pulmonary
interstitial emphysema of RDS. Inten Care Med 1997;23(6): 671
676.
30. Goddard P, et al. Use of continuously recording intravascular
oxygen electrode in the newborn. Arch Dis Child 1974;49(11):
853860.
31. Weiss IK, Fink S, Harrison R, Feldman JD, Brill JE. Clinical
use of continuous arterial blood gas monitoring in the pediatric intensive care unit. Pediatrics 1999;103:440445.
32. Morgan C, et al. Continuous neonatal blood gas monitoring
using a multiparameter intra-arterial sensor. Arch Dis Child
Fetal Neonatal Ed 80(2):F93F98.
33. Rais-Bahrami K, Rivera O, Mikesell GT, Short BL. Continuous blood gas monitoring using in-dwelling optode method:
Comparison to intermittent arterial blood gas sampling in
ECMO patients. J Perinatol 2002;22(6):472474.
34. Widness JA, et al. Clinical performance of an in-line pointof-care monitor in neonates. Pediatrics 2000;106(3): 497 504.
35. Gardner RM. Invasive pressure monitoring. In: Civetta JM,
Taylor RW, Kirby RR. editors Critical Care, 3rd ed. Philadelphia: Lippincott-Raven; 1997. pp 839845.
36. http://www.utahmed.com/deltran.htm.
37. Ewert P, Nagdyman N, Fischer T, Gortner L, Lange PE.
Continuous monitoring of cardiac output in neonates using
an intra-aortic Doppler probe. Cardiol Young 1999;9(1):
4248.
38. Tsai-Goodman B, et al. Development of a system to record
cardiac output continuously in the newborn. Pediatr Res.
1999;46(5):621625.
39. Hermansen MC, Hermansen MG. Intravascular catheter
complications in the neonatal intensive care unit. Clin Perinatol; 2005;32(1):141156.
40. Barrington KJ. Umbilical artery catheters in the newborn:
Effects of position of the catheter tip. Cochrane Database Syst
Rev CD000505, 2000.
41. Kempley ST, Bennett S, Loftus BG. Randomized trial of
umbilical arterial catheter position: Clinical outcome. Acta
Paediatr 1993;82:173176.
42. Mokrohisky ST, Levine RL, Blumhagen JD. Low positioning
of umbilical-artery catheters increases associated complications in newborn infants. N Engl J Med 1978;299:561564.
43. Umbilical Artery Catheter Trial Study Group. Relationship
of intraventricular hemorrhage or death with the level of
umbilical artery catheter placement: A multicenter randomized clinical trial. Pediatrics 1992;90:881887.
44. Haldeman S, Fowler GW, Ashwal S. Acute flaccid neonatal
paraplegia: A case report. Neurology 1983;33:9395.
45. Cumming WA, Burchfield DJ. Accidental catheterization of
internal iliac artery branches: A serious complication of umbilical artery catheterization. J Perinatol 1994;14:304309.
46. Klaus MH, Fanaroff AA. Care of the high-risk neonate, 5th
ed. Philadelphia: WB Saunders; 2001.
47. Nowlen TT, Rosenthal GL, Johnson GL. Pericardial effusion
and tamponade in infants with central catheters. Pediatrics
2002;110:137142.
48. OGrady NP, et al. Guidelines for the prevention of intravascular catheter-related infections. Pediatrics 2002;110(5):e51.
49. Tudos AJ, Besselink GAJ, Schasfoort RBM. Trends in miniaturized total analysis systems for point-of-care testing in
clinical chemistry. Lab on a Chip: 2001;1(2):8395.
50. Wolfbeis OS. Fiber-optic chemical sensors and biosensors.
Anal Chem 2001;74(12):26632677.
51. Zhang XC. Terahertz wave imaging: Horizons and hurdles.
Proceedings of the First International Conference on Biomedical Imaging and Sensing Applications of THz Technology,
Physics in Medicine Biology 2002;47(21):36673677.
See also ARTERIES,
ELASTIC PROPERTIES OF; BLOOD GAS MEASUREMENTS;
FIBER OPTICS IN MEDICINE; NEONATAL MONITORING; STRAIN GAGE.
597
and assays employed in the diagnosis and treatment of

diseases as well as their utility in the research laboratory.
THE IMMUNE SYSTEM AS IT APPLIES TO ANTIBODIES (1)
The immune system is normally directed at foreign molecules borne by pathogenic microorganisms. However, the
immune system can also be induced to respond to simple
nonliving molecules. Any substance that can elicit an
immune response is said to be immunogenic and is called
an immunogen. There is a clear operational distinction
between an immunogen and an antigen. An antigen is
defined as any substance that can bind to a specific antibody (see below), but is not necessarily able to elicit an
immune response by itself.
Immunization
The deliberate induction of an immune response is known
as immunization. To determine whether an immune
response has occurred and to follow its course, the immunized individual is usually monitored for the appearance of
antibodies directed at the specific antigen. Monitoring the
antibody response usually involves the analysis of relatively crude preparations of sera. The serum is the fluid
phase of clotted blood, which contains a variety of specific
antibodies against the immunizing antigen as well as other
soluble serum proteins.
Cells Participating in an Immune Response
B lymphocytes (or simply B cells) are one of the two major
types of lymphocytes that enable the adaptive immune
response. When activated, B cells differentiate into plasma
cells that secrete antibodies. T lymphocytes or T cells
consist of three main classes. One class differentiates upon
activation into cytotoxic T cells, which may kill foreign
tissues, cancer cells, and cells infected with virus. The
second class of T lymphocytes is T helper cells that differentiate into cells that activate and enable the proper
function of other cells, such as B cells. The third class is
the T suppressor cells that limit the extent of the immune
response.
Antigen Recognition
BRENDA H. LASTER
JACOB GOPAS
Ben Gurion University of the
Negev
Beer Sheva, Israel
INTRODUCTION
This article outlines the association of antibodies within
the human immune system, the structural and binding
characteristics of antibodies, and the development and
production of monoclonal antibodies. Recent advancements in recombinant DNA techniques and genetic engineering are described, including the use of plants to
increase the production capacity of Mabs. Their usefulness
as biological and medical reagents is further elaborated in
a description of the various instrumentation, techniques,
Both T and B lymphocytes bear receptor proteins on their

surface that allow them to recognize antigen. Collectively,
these receptors are highly diverse in their antigen specificity, but each individual lymphocyte is equipped with
membrane-bound receptors that will recognize only one
particular antigen. Each lymphocyte therefore recognizes a
different antigen. Together, the receptors of all the different lymphocytes are capable of recognizing a very wide
diversity of antigens, which encompass most of the different antigens an individual will meet in a lifetime. These
include those antigens that are exclusively synthesized in
the laboratory. The B cells do not secrete antibody until
they have been stimulated by specific antigen. The B-cell
antigen receptor (BCR) is a membrane-bound form of the
same antibody that they will secrete when activated by
antigen. Thus the antigen recognized by both the BCR and
598
The antigen-binding region varies extensively among

antibody molecules and is thus known as the variable
region or V region, labeled as VH (heavy chain) and VL
(light chain). It is this variability that allows each antibody
molecule to recognize and bind a particular antigen. The
total repertoire of antibodies made by a single individual is
large enough to ensure that virtually any structure can be
bound. The association between the antibody and the
antigen depends on their steric conformation. That is,
depending on the size and interatomic distance of these
reacting molecules, a tight fit between the antibody combining sites and the antigenic determinant can occur.
The region of the antibody molecule that engages the
effector functions of the immune system, but is not associated with antibody binding and does not vary in the same
way is known as the constant region or C region. It is
typified by the IgG antibody shown in Fig. 1 and is designated CL (light chain) and CH (heavy chain). It has five
main forms, or isotypes, that are specialized for activating
the different immune effector mechanisms.
The remarkable diversity of antibody molecules is the
consequence of a highly specialized mechanism by which
the secreted antibody present in the same B cell, are

identical.
Antibodies
Antibody molecules as a class are now generally known as
immunoglobulins (Ig), and the antigen receptor of B lymphocytes is known as surface immunoglobulin. The T cell
antigen receptor (TCR) is related to immunoglobulins, but
is quite distinct from it in structure and function.
Structure and Function of Antibodies
Antibodies are the antigen-specific products (proteins)
secreted by B cells. The antibody molecule has two separate
functions: one is to bind specifically to molecules from the
immunogen (pathogen) that elicited the immune response;
the other is to recruit various cells and molecules in order
to remove and destroy the pathogen once the antibody is
bound to it. These functions are structurally separated in
the antibody molecule. One region of the antibody specifically recognizes antigen and the other engages the effector
mechanisms that will dispose of it.
Antigen
binding
NH
NH
VH
NH
VH
VL
Pepsin Cleavage
F (ab')z fragment
Antigen
binding
NH
CH1
CH2
Fab fragment
CL
CH2
Carbohydrate
O
CO
lostype Determinate
,,,, or
s s
s s
CL
VL
CH1
CH3
CH3
COOH
COOH
Papain
Cleavage
CO
OH
Fc fragment
Heavy chain constant region.

Heavy chain variable region.
Light chain constant region.
Light chain variable region.
Carbohydrate.
Antigen binding site.
s s
Disul fide bonds.

Papain cleavage site.
Pepsin cleavage site.
Figure 1. Basic immunoglobulin structure. [Courtesy of Sigma-Aldrich (www.sigmaaldrich.com/

img/assets/8181/AntibodyExp).]
the genes that code for antibody production and that are
expressed in any given B cell are assembled by DNA
rearrangements that join together two or three different
segments to form a V region gene during the development
of the B cell. Subsequent DNA rearrangement can attach
the assembled V-region to any C-region gene and thus
produce antibodies of any of the five isotypes.
Antibodies are roughly Y-shaped molecules. All antibodies are constructed in the same way from paired heavy
and light polypeptide chains. In Fig. 1, one can observe that
the innermost regions of the Y-shaped molecule are the
heavy chains; the light chains are the outermost regions.
Within this general category, however, five classes (isotypes) of immunoglobulin -IgM, IgD, IgG, IgA, and IgE- can
be distinguished biochemically as well as functionally. The
five classes are defined by the structure of their heavy
chain. Their distinctive functional properties are conferred
by differences in the amino acid sequences of the carboxyterminal part of the heavy chain (COOH in Fig. 1) in the
region that is not associated with the light chain. IgG
(Fig. 1) will be used to describe the general structural
features of immunoglobulin molecules.
The IgG antibodies are large molecules (150 kDa)
composed of two different polypeptide chains. One of these
polypeptide chains, 50 kDa in size, is termed the heavy or
H chain. The other, is 25 kDa in size, and is termed the
light chain or L chain. The two chains are present in an
equimolar ratio, and each IgG molecule contains two heavy
chains and two light chains. The two heavy chains are
linked to each other by disulfide bonds and each heavy
chain is linked to a light chain by a disulfide bond. In any
one immunoglobulin molecule, the two heavy chains and
the two light chains are identical, enabling them to bind
two identical antigenic determinants.
The amino-terminal sequences of both the heavy and
light chains vary greatly among different antibodies. The
variability in sequence is limited to approximately the first
110 amino acids on the chain, corresponding to the first
domain, whereas the carboxy-terminal sequences are constant between immunoglobulin chains, either light or
heavy, of the same isotype.
599
to as humanization. This avoids their recognition as

foreign, and prevents their rapid clearance from the body.
The process utilizes the variable region of a mouse antibody
coupled to the Fc region from human antibodies. Antibody
fragments may also be coupled to toxins, radioactive isotopes and protein domains that interact with effector
molecules or cells.
Antibody Heterogeneity
The antibodies generated in a natural immune response or
after immunization in the laboratory are a mixture of
molecules of different antigen specificities and affinities.
Because of their multiple specificities they are termed
polyclonal antibodies. Some of this heterogeneity results
from the production of antibodies that bind numerous
different antigenic determinants (epitopes) present on
the immunizing antigen. However, even antibodies directed at a single antigenic determinant can be markedly
heterogeneous. Antisera (serum containing antibodies
against specified antigens) are valuable for many biological
purposes, but they have certain inherent disadvantages
that relate to the heterogeneity of the antibodies they
contain. First, each antiserum is different from all other
antisera, even when raised in a genetically identical animal while using the identical preparation of antigen and
immunization protocol. Second, antisera can be produced
in only limited volumes, and thus it is impossible to use the
identical serological reagent in a long or complex series of
experiments, or in clinical tests or therapy. Finally, even
purified antibodies may include minor populations of antibodies that give unexpected cross-reactions that confound
the analysis of experiments and can be harmful in therapy.
To avoid these problems, and to harness the full potential of
antibodies, it became necessary to develop a method for
making an unlimited supply of antibody molecules of
homogeneous structure and known specificity. This has
been achieved through the production of monoclonal antibodies from hybrid antibody forming cells or, more
recently, by genetic engineering.
Fragmentation of Antibodies
The antibody molecule can be readily cleaved by different
proteases into functionally distinct fragments. For example, Fab fragments, each of which consists of two identical
fragments, each containing the antigen binding region.
Additionally, an Fc fragment can be extracted that interacts with effector molecules and cells, or one F(ab0 )2 fragment that contains both arms of the antigen binding
region. Figure 1 shows the sites of the derivation of these
fragments. Genetic engineering techniques now permit the
construction of designed variations of the antibody molecule such as a truncated Fab that comprises only the V
region of a heavy chain linked to a V region of a light chain.
This is called a singlechain Fv. A broad range of genetically engineered molecules are now becoming valuable
therapeutic agents because their smaller size readily permits their penetration into tissue. Useful antibodies from
animal sources have been engineered in a process referred
Production of Monoclonal Antibodies

In 1975, using cell culture techniques, Kohler and Milstein
(2) found a way to grow an immortal B-cell-like lymphocyte
that continuously produced an antibody with a predetermined specificity. The procedure required the immunization of a mouse in order to produce a large population of
B-cells in the spleen that would secrete a specific antibody.
However, in cell culture, the life span of spleen cells is only
a few days. To produce a continuous source of antibody,
the B cells had to grow continuously. This was achieved by
fusing the spleen cells that contained a particular gene,
the hypoxanthine-guanine phosphoribosyl transferase
(HGPRT) gene with immortal myeloma cells (cancerous
plasma cells). In general, plasma cells are mature-antibody
secreting B cells. The myeloma cells were preselected to
ensure three specific properties. First, immortality in cell
culture; second, that they were sufficiently altered so that
600
permitted only hybrid cells to survive in the HAT medium,

because only hybrid cells would be able to grow in the
culture due to the conferral of immortality by the myeloma
cells. Therefore, the immune spleen cells conferred both
antibody specificity and the HGPRT gene to the hybrid cell,
while the myeloma cell conferred immortality to the spleen
cell and they were able to survive indefinitely in culture.
This is the method that is used today to produce individual
hybridomas (the hybrid cells) that are then screened for
antibody production. Single antibody-producing cells that
produce an antibody with the desired specificity are cloned.
These cloned hybridoma cells are grown in bulk culture to
produce large amounts of antibody that are used in a
variety of ways. Since each hybridoma is descended from
a single B cell, all cells of a particular hybridoma cell line
produce the same hybridoma molecule. This is the monoclonal antibody (Mab). The procedure is shown as a schematic outline in Fig. 2.
Antigen
Myeloma cells
(HGPRT-)
Fusion
Spleen cells
1.Culture in
HAT medium
2.Test each
supernatant
for antibodies
3. Clone each
positive culture
4. Test each
supernatant
for antibodies
5. Expand positive clones
propagate
in vitro
in vivo
or
Harvest monoclonal antibodies
Figure 2. Schematic of hybridoma protocol. [Courtesy of

Prof. John Kimball (http://users.rcn.com/jkimball.ma.ultranet/
Biology Pages/M/Monoclonals.html).]
they did not secrete antibody, and third, that they would
not flourish in a particular growth medium containing
hypoxanthine, aminopterin and thymidine (HAT). The
HAT medium was previously shown to be a highly selective
medium for those specific hybrid cells that lack the gene for
the enzyme, HGPRT. Consequently, all unfused myeloma
and spleen cells would not survive in the HAT medium. The
HGPRT gene that was contributed by the spleen cell
Figure 3. Antibody engineering: examples of antibody-based constructs

and respective routes for their generation. Hybridoma technology provides a
way of producing mouse monoclonal
antibodies. Genetic engineering has
encouraged the creation of chimeric,
humanized, and human antibodies,
antibody fragments (traditionally
obtained by partial digestion of
immunoglobulins with proteases),
multimeric antibody fragments, fusion
(immunoadhesins and immunotoxins)
and bispecific antibodies. Multimeric
antibody fragments (diabody, triabody, and tetrabody) are represented
as multivalent structures, although
they can also be engineered to be
multispecific. The minibody depicted
is a dimer that can be linked to the
CH3 fragment via a LD linker or a
flexible linker (FlexMinibody). Bispecific F(ab)2 is shown as a Fab dimer
linked noncovalently via interaction
of amphipathic helices. (Courtesy
of American Chemical Society and
American Institute of Chemical Engineers, Copyright # 2004.) (See Ref. 3.)
Recombinant Genetic Engineering

The field of antibody engineering endeavors to improve the
target specificity and effector function of the antibodies by
altering their construction, while retaining their binding
characteristics. This is achieved by using new recombinant
engineering technologies to redesign the molecular architecture of the antibodies. Examples of this are shown in
Fig. 3, where in a process called genetic engineering,
recombinant DNA (spliced DNA that is formed from two
or more sources and then rejoined) is produced by putting
genes from one species into the cells of an individual
organism of another species. The foreign DNA becomes
Mouse
CDR-Grafted Chimeric
Mouse
Hybridoma
Human
Enzyme
digestion
(papain, pepsin)
Transgenic
mouse
In vitro
Ab
libraries
scFv
pFc'
Fc
Fab
Chemical
conjugation
Recombinant
systems
F(ab')2
F(ab')2
Immunoadhesin
Bispecific F(ab)2
Fab
Fv
Immunotoxin
Diabody
scFv2
Triabody
Minibody
part of the hosts genome (its genetic content) and is

replicated in subsequent generations descended from that
host. An alternative technique for producing antibody-like
molecules is the Phage Display Libraries for Antibody
V-region Production (4). In this approach, gene segments
that encode the antigen-binding variable region of antibodies are fused to genes that encode the coat protein (outside surface) of a bacteriophage (viruses that infect bacteria).
In essence, mRNA from primed human B-cells is converted
to cDNA. The large variety of diverse antibody genes are
expanded by the polymerase chain reaction (PCR) to generate a highly diverse library of antibody genes. Bacteriophage containing such gene fusions are then used to
infect bacteria, resulting in phage particles that have outer
surfaces that express the both antibody-like fusion protein,
and the same antigen-binding domain displayed on the
outside of the bacteriophage. The collection of such recombinant phages, each displaying a different antigen-binding
domain on its surface, is known as a phage display library.
A particular phage can be isolated from the mixture and
can be used to infect fresh bacteria. Each phage isolated in
this way produces a monoclonal antigen-binding particle
analogous to a monoclonal antibody. A complete antibody
molecule can then be produced by fusing the V region of a
particular phage with the invariant part of the immunoglobulin gene. These reconstructed antibody genes can be
introduced (transfected) into a suitable host cell line. These
genes will become part of the cells genome and will secrete
antibodies akin to hybridomas.
Monoclonal Antibodies Produced in Plants, Plantibodies
After undergoing genetic engineering techniques, plant
cells are capable of assembling and producing unlimited
quantities of antibodies, referred to as plantibodies (5).
This trademark name for human antibodies manufactured
in plants has functionally limitless production capacity and
lower costs than those associated with the yeast fermentation process that is currently being used to produce mass
quantities of human antibodies. This fairly recent finding
might prove to be of benefit in the medical, consumer, and
industrial applications of monoclonals. For example, it has
been postulated that the development of plantibodies with
a capability of sequestering heavy metals or radioactive
compounds might have a very positive impact on the
environment, particularly because their production is very
inexpensive and large supplies are easily produced.
Because the corn crop is so readily available worldwide,
and its kernel stores natural plantibodies, these can be
purified as needed by standard milling procedures. Potato
and tomato crops are also being used. The first clinical use
of the effectiveness of a plantibody was against the bacterium, Streptococcus mutans. This organism produces lactic
acid that erodes tooth enamel. The plantibody was brushed
onto human teeth for 3 weeks and tooth decay was prevented for up to 4 months. The action of the antibody was to
prevent the bacterium from binding to the tooth surface.
Plantibody-containing gels are being developed to prevent
genital herpes infections and to protect newborn babies
during delivery against transmission of the Herpes virus
from infected mothers. Plantibodies against human immu-
601
nodeficiency verus (HIV) and the production of sperm are

also being developed. Concerns have been expressed about
the use of genetically engineered food crops because of the
potential dangers of their getting into the wrong hands, or
disturbing the ecological balance.
The aforementioned technologies have revolutionized
the use of antibodies by providing a limitless supply of
antibodies with single and known specificity. Monoclonal
antibodies are now used in most serological assays, as
diagnostic probes and as therapeutic agents.
TECHNIQUES FOR USING MONOCLONAL ANTIBODIES
AS SEROLOGICAL AND DIAGNOSTIC PROBES
Monoclonal antibodies can serve as tools for diagnosing
and treating disease and are valuable agents in the
research laboratory. Their utility required the development of procedures that would permit them to be viewed
at the particular region of interest. Some of the most widely
used techniques are described in the following sections (1).
Immunofluorescence
Since antibodies bind stably and specifically to their corresponding antigen, they are invaluable as probes for
identifying a particular molecule in cells, tissues, or biological fluids. Monoclonal antibody molecules (Mabs) can
be used to locate their target molecules accurately in single
cells or tissue sections by a variety of different labeling
techniques. When either the antibody itself, or the antiMab that is used to detect it, is labeled with a fluorescent
dye the technique is known as immunofluorescence. As in
all serological techniques, the antibody binds stably to its
antigen, allowing any unbound antibody to be removed by
thorough washing. The fluorescent dye can be covalently
attached directly to the specific antibody, but more commonly, the bound antibody is detected by a secondary
fluorescent anti-immunoglobulin; that is, the first antibody
binds to the antigen and a fluorescent secondary antibody
(antibody) is targeted to the primary antibodyantigen
complex. The technique is known as indirect immunofluorescence, which is demonstrated in Fig. 4, where the binding
of the first antibody to the antigen is followed by the
binding of the antibody. The dyes chosen for immunofluorescence are excited by light of a particular wavelength, and
emit light of a different wavelength in the visible spectrum.
By using selective filters that can permit only certain
wavelengths of light to pass, only that light coming from
the dye or fluorochrome used is detected in the fluorescence
microscope. Therefore, the antibody can be located by
virtue of its emission of fluorescent light. The recently
developed confocal fluorescent microscope considerably
enhances the resolution of the technique. If different dyes
are attached to different antibodies, the distribution of two
or more Mabs can be determined in the same cell or tissue
section. Differentiating between the antibodies occurs
because either the dye or the fluorochrome will excite at
different wavelengths or because they will emit their fluorescence at different wavelengths. An example of the immunofluorescence technique is shown in Fig. 4, whereby,
through the use of monoclonal antibodies targeted to
Solid
602
Add antigen
Antibody
Add antibody-enzyme conjugate
Enzyme
Add substrate
Substrate
(colorless)
Product
Figure 4. Indirect immunofluorescence. The primary antibody
binds to the antigen and the fluorescent antiantibody binds to
the primary thereby increasing the signal. [Courtesy of Prof. v.
Sengbusch (http://www.biologie.uni-hamburg.de/b-online/d00/
copyrig.htm).]
intracellular proteins, certain structures within the cell

become visible. The structure shown in Fig. 5 is that of the
spindle, which appears during cell division. During certain
phases of cell division, the chromosomes arrange themselves
in the equatorial plane of the spindle. The spindle is made
up of microtubules that, in turn, are composed of proteins.
Monoclonal antibodies that would bind to two specific proteins, a and g-tubulin, of the microtubule were synthesized
and labeled with two different fluorochromes. During cell
division, the cells were exposed to the fluorescent-labeled
Mabs that formed a complex with the proteins and permitted visualization of the spindle.
Immunohistochemistry
An alternative to immunofluorescence for detecting a protein in tissue sections is immunohistochemistry, in which
the specific Mab is chemically coupled to an enzyme that
converts a colorless substrate into a colored reaction pro-
Figure 6. Schematic of ELISA assay protocol. [Courtesy of Prof.

John Kimball (http://users.rcn.com/jkimball.ma.ultranet/Biology
Pages/E/Elisa.html).]
duct in situ. The Enzyme-Linked ImmunoSorbent Assay

(ELISA) is a technique that detects and quantifies specific
antigens from a mixture. It is widely used in procedures
that screen blood for viral or bacterial contamination, to
detect infections, toxins, illegal drugs, or allergens, and in
measuring hormone levels, such as in pregnancy or thyroid
function. The assay involves the binding of an antibody to a
solid surface and exposing it to the antigens. A second
complex, consisting of the same antibody, but additionally
tagged with a particular enzyme, is exposed to the initial
antibodyantigen conjugate and binds. After washing the
surface to remove excess unbound antigen, a colorless
substrate is added that permits the antigen to be converted
into a colored product that can be read and measured using
absorption spectrometry. The intensity of color is proportional to the concentration of bound antigen. A schematic of the ELISA assay is shown in Fig. 6. A more
detailed description of the Elisa procedure can be found
in (Kimballs Biology Pages http://biology-pages.info).
Horseradish peroxidase and alkaline phosphatase are also
used as enzymes in immunochemistry assays. An example
of the utility of these enzymes for protein detection is
shown in Fig. 7.
Immunoelectron Microscopy
Figure 5. Dividing cells (mitosis: metaphase, anaphase, and

telophase) were stained with monoclonal antibodies against two
intracellular proteins, a-tubulin in green, and g-tubulin in red.
Because these proteins constitute the spindle, the intracellular
structure upon which chromosomes line up during mitosis, the structure is visualized by virtue of the difference in the fluorochromes tagged to the Mabs that were bound to the proteins.
Chromosomes were stained with a blue dye. (www.img.cas.cz/dbc/
gallery.htm.)
Antibodies can be used to detect the intracellular location

of structures or particular molecules by electron microscopy, a technique known as immunoelectron microscopy.
After labeling Mabs with gold particles and targeting them
to samples, they can then be examined in the transmission
electron microscope. Since electrons do not penetrate
through gold particles, the regions in which the antibodies
bind appear as dark dots.
603
bound to the antibodies are revealed by enzyme-labeled, antiimmunoglobulin antibodies. By this technique the presence
or absence, as well as the amounts of specific proteins, can be
monitored following a variety cell treatments. Specific DNA
labeled with antigen (hapten)-bound nucleotides can be
blotted onto a membrane and detected with Mabs against
the hapten. This allows the detection of viral or bacterial
DNA in tissues or body fluids, as generated by PCR.
Purification Techniques
Figure 7. Detection of the measles virus (MV) P-protein by

Western Blot in MV infected and noninfected cells. Whole-cell
lysates were prepared from MV that was either persistently
infected (NS20-MV) or notinfected (NS20) mouse neuroblastoma
cells. The proteins in the lysates were separated by SDSPAGE
and blotted onto nitrocellulose paper. The blot was incubated with
a Mab against the MV P-protein, followed by a secondary
antimouse immunoglobulin antibody linked to horseradish
peroxidase. The P-protein band was detected when a substrate
was added that was modified by peroxidase on the blot and caused
light to be released. Light was detected on a specific band after
exposure to film. The results show that only the measles- infected
cells express the viral protein. (Courtesy of Jacob Gopas.)
Blotting Techniques
Immunoblotting or Western blotting is used to identify the
presence of a given protein in a cell lysate. Cells are placed
in detergent to solubilize all cell proteins and the lysate
(the material resulting from the ruptured cells) is run on
sodium dodecyl sulfate polyacrylamide gel electrophoresis
(SDSPAGE), which enables protein migration and
separation by size. Further resolution is achieved if the
proteins are initially separated by charge according to their
isoelectric point and then by size. This technique is referred
to as two-dimensional (2D) gel electrophoresis. The proteins are then transferred (blotted) from the gel to a stable
support, such as a nitrocellulose membrane for easier
handling. Specific proteins of interest in the lysates mixture are detected by incubating the membrane with a Mab
that can react with a defined protein on the membrane. An
example of the technique is shown in Fig. 7. The proteins
Affinity chromatography and immunoprecipitation are

techniques that enable purification of molecules and their
characterization. A mixture of molecules can be incubated
with a Mab, which is chemically attached to a solid support.
The bound antibodyantigen complex is washed from
unbound molecules by centrifugation, and then the molecule of interest is eluted for further characterization. These
techniques are useful for protein purification, for determining its molecular weight, its abundance, distribution, and
whether it undergoes chemical modifications as a result of
processing within the cell.
Immunoelectrophoresis
Two-dimensional electrophoresis is used to separate different antigens that might be present in one solution. The
antigens are separated on the basis of their electrophoretic
mobility. The currents are run at right angles to each other,
driving the antigens into the antiserum (containing Mabs).
Peaks are obtained when the antigen forms a complex with
the antibody; the area under the peaks gives the concentration of antigen as shown in Fig. 8. Rocket electrophoresis is a similar technique. Here, after a current is applied,
the antigens are separated based upon their ionic charge by
their differential migration through a gel that contains
antibody. As shown in Fig. 9, concentration is determined
by the migration distance. In countercurrent electrophoresis, the greater internal osmotic pressure drives the antibody backwards into a gel after a current is applied. An
antigen that is negatively charged will form a complex with
the antibody in the gel in a pH-dependent process.
Instrumentation
An immensely powerful tool for defining and enumerating
and isolating cells is the use of the fluorescence-activated
Precipitin arc
+
Antibody in gel
Ag2
Ag1
First run
Starting
well
Second run
Figure 8. Two-dimensional immunoelectrophoresis. Antigens are separated on the basis of electrophoretic

mobility. [Courtesy of the Natural
Toxins Research Center at Texas
A&M University Kingsville (http://
ntri.tamuk.edu/).]
Antibody
in agarose
gel
Precipitin
arcs
(rockets)
pH 8.6
101
Antigen
wells
104
CD19 PE
103
102
604
Figure 9. Rocket electrophoresis. Antigen is electrophoresed into

gel containing antibody. The distance from the starting well to the
front of the rocket shaped arc is related to antigen concentration.
[Courtesy of the Natural Toxins Research Center at Texas A&M
University Kingsville (http://ntri.tamuk.edu/).]
cell sorter (FACS). This instrument is used to study the

properties of cell subsets identified using Mabs to cell
surface proteins. Individual cells are first tagged by treatment with specific fluorescent Mabs. The mixture of
labeled cells is then forced with a much larger volume of
fluid through a nozzle, creating a fine stream of liquid
containing cells spaced singly at intervals. As each cell
passes through a laser beam it scatters the laser light, and
any dye molecules bound to the cell will be excited and
fluoresce. Sensitive photomultiplier tubes detect both the
scattered light, which gives information on the size and
granularity of the cell, and the fluorescence emission,
provide quantification of the binding of the labeled Mabs,
and on the expression of cell-surface proteins by each cell.
In the cell sorter, the signals passed back to the computer
are used to generate an electric charge, which is passed
from the nozzle through the liquid stream. Droplets containing a charge can then be deflected from the main
stream as they pass between plates of opposite charge.
In this way a specific population of cells, distinguished by
the binding of the labeled antibody and its defined electrical charge, can be extracted and purified from a mixed
population of cells. Alternatively, to deplete a population of
cells, a labeled antibody directed at marker proteins
expressed by undesired cells will direct the cells to a waste
channel, retaining only the unlabeled cells. Several Mabs
labeled with different fluorochromes can be used simultaneously. FACS analysis can give quantitative data on the
percentage of cells bearing different molecules, and the
relative abundance of the particular molecules in the cell
population, 10,000 cells in a typical experiment demonstrates the retrieval of data after FACS analysis. An example of data output from FACS is shown in Fig. 10.
Mabs as Molecular Probes
Mabs can also be used to determine the function of molecules. Some antibodies are able to act as agonists, when the
binding of the Mab to the molecule mimics the binding of
the natural ligand (antigen) and activates its function. For
example, antibodies to the CD3 antigen present on mature
human T cells have been used to stimulate the T cells. This
100
Increasing antigen
concentration
100
101
102
CD3 FITC
103
104
Figure 10. FACS analysis. Characterizing cells at different

stages of development through the use of fluorescent labeled
monoclonal antibodies against cell surface markers is one of the
most common applications of flow cytometry. Changes in the
relative numbers, absolute counts, or in the ratio of cell types
can provide valuable information as to the status of the immune
system in human disorders or animal models. Different cell types
can be detected and quantified from a mixed population by the use
of monoclonal antibodies labeled with different fluorescent dyes
that have nonoverlapping emission spectra. In this example
experiment, blood lymphocytes were incubated with two
different Mabs, CD19, and CD3. CD19 was labeled with the
fluorochrome phycoerythrin (PE) and binds a cell membrane
molecule specific for B-lymphocytes. The CD3 was labeled with
fluorecein isothyocyanate (FITC) that detects a cell membrane
protein specific for T-lymphocytes. Three populations of cells
were detected in this experiment according to the antibody
bound to the cells. The logarithmic x and y axis represent
relative amounts of fluorescence detected on cells labeled with
FITC or PE, respectively. The blue dots represent cells unstained
by either of the antibodies, the red dots represent B-lymphocytes
that were detected by CD19 and the green dots represent
T-lymphocytes, CD3 positive cells. No cells were detected that
bound both antibodies (top right quadrant).
occurs because CD3 is associated with the T-cell receptor

and is responsible for signal transduction of the receptor.
Conversely, Mabs can function as antagonists, inhibiting
the binding of the natural ligand and thus blocking its
function. For example, antibodies that block the epidermal
growth factor receptor (a growth stimulating protein) function as antagonists.
THE USE OF MONOCLONAL ANTIBODIES
AS THERAPEUTIC AGENTS
Mabs against cell-surface molecules have been used to
remove specific lymphocyte subsets or to inhibit cell function in vitro. Cytotoxic drugs kill proliferating cells indiscriminately. In contrast, antibodies can interfere with
immune responses in a nontoxic and much more specific
manner. For example, Mabs can be used to remove undesirable lymphocytes from donor bone marrow cells prior to
transplantation. This treatment selectively removes lymphocytes that recognize the host tissues as foreign and
induce a potentially fatal condition known as Graft versus
Host reaction (6).
Mabs are being tested experimentally to inhibit transplant rejection, to alleviate and suppress autoimmune
disease and in cancer detection and treatment. The major
impediment to therapy with monoclonal antibodies in
humans is that these antibodies are mostly of mouse origin,
and humans rapidly develop antibody responses to mouse
antibodies. This not only blocks the actions of the mouse
antibodies, but leads to allergic reactions. If this occurs,
future treatment of the same patient with any mouse Mab
is unacceptable. In principle, the problem can be avoided by
producing antibodies that are not recognized as foreign by
the human immune system. Several strategies are being
explored for their construction. One approach is to clone
human V regions into a phage display library (see above)
and select for its ability to bind human cells. With this
method, Mabs that are entirely human in origin can be
obtained. Second, mice that lack endogenous immunoglobulin genes can be made transgenic (chimeric). That is,
they can have human genes put into their genome through
recombinant DNA techniques. When this occurs, they will
then express human immunoglobulin heavy and light
genes and eventually antibody molecules. A third approach
is to graft the variable region of a mouse Mab into the rest
of the human immunoglobulin molecule, in a process
known as humanization. These recombinant antibodies
are far less immunogenic in humans than the Mabs of
the parent mouse, therefore, they can be used for more
efficient and repeated treatment of humans with far less
risks. In some cases even humanized antibodies may evoke
an immune response and must be administered with
immunosuppressive drugs.
THE USE OF MONOCLONAL ANTIBODIES IN THE
DETECTION, FOLLOW-UP, AND TREATMENT OF CANCER
Tumor-Specific Antigens
For the greater part of the twentieth century, it was
assumed that any antigens present on the cell surface of
tumor cells would also be present in normal cells; therefore,
few investigations were undertaken to elicit any autoimmune response against cancer cells. However, once inbred
mouse strains bearing transplanted syngeneic (genetically
identical) tumors became available, research studies validated that immune reactions against these tumors could be
induced with no toxic effects on normal tissues, and scientists began to pursue the identification of tumor specific
antigens. Shared tumor antigens were found in many of
the same types of cancers in different patients, and unique
antigens were isolated that were specific for a particular
cancer in a particular patient. The SEREX database lists
the antigens that have been isolated from humans (7).
These antigens have the ability to generate an immune
response when introduced into a patient.
The advent of monoclonal antibodies suggested the
possibility of targeting and destroying tumors by making
antibodies against tumor-specific antigens. However, this
relies upon the identification of a tumor-specific antigen
that is located on the surface of cancer cells. Because of
their ability to differentiate between normal and malignant tissues and to exact a variety of antitumor responses,
605
Mabs offer a significant advantage to conventional forms of

therapy. Several monoclonal antibodies have already been
proven to be relatively well tolerated and effective for the
treatment of many different malignant diseases.
Approaches to Cancer Immunotherapy
Approaches to cancer immunotherapy can be either active
or passive. For example, in the active category, tumor
vaccines that immunize against particularly defined tumor
antigens, can be used. In the passive category is the use of
monoclonal antibodies that are either conjugated, unconjugated, or radiolabeled. These same approaches can also
be categorized as specific, wherein antigens are directly
targeted, or nonspecific, where immune cells are used to
directly target tumor cells. Other approaches are taken
that elicit antitumor effects with different mechanisms,
such as using antibodies to block growth factors or receptors on cells; targeting specific tissue components of the
tumor or its blood vessels; interfering with cell signals; or
with apoptosis (programmed cell death) (8).
Magic Bullets
While such Mab-based therapies offer a high potential to
fulfill the promise of magic bullets for the treatment of
malignant disease, successful application of these therapies is often impaired by several impediments. Factors
inhibiting the therapeutic benefit of Mabs may include
low or heterogeneous expression of target antigens by
tumor cells, high background expression of antigen on
normal cells, host antibody immune responses to the Mabs
themselves, insufficient anti-tumor response after Mab
binding, as well as physical obstructions preventing antibody binding, such as crossing to and from blood vessels as
well as tissue barriers en route to the solid tumor mass (9).
These factors influence the ability of the Mabs to penetrate
to the tumor.
IMAGING TUMORS WITH MONOCLONAL ANTIBODIES
Mabs in Nuclear Medicine
The presence of malignant tumors can be detected through
the use of monoclonal antibodies radiolabeled most frequently with the isotopes technetium-99m (99mTc) or
indium-111 (111In). The particular label selected depends
upon the size of the antibody. For example, large fragments
or whole antibodies require a longer half-life isotope, such
as 111In (T1/2 2.8 days), whereas smaller Fab fragments,
that are cleared from the body more quickly, can be labeled
with 99mTc (T1/2 6 h). Imaging is performed by a Single
Photon Emission Computed Tomography (SPECT) camera
whose detectors scan the body and register the radioactive
counts. The counts are then mathematically transformed
into an image that displays the sites of radioactivity. The
nuclear medicine procedure that utilizes this procedure is
known as Tumor-Specific Monoclonal Antibody Radioscintigraphy. Because of occasional difficulties with these
techniques, such as inadequate tumor perfusion, inadequate amounts of antigen on the surface of the tumor
cells, antigen heterogeneity, and nonspecific uptake, new
606
approaches are being investigated. However, due to limited

clinical experience, it is too early to predict whether they
will improve imaging performance (10). Among these
methods is the use of other imaging techniques, such as
bone scans or computed tomography (CT), in conjunction
with SPECT. In other approaches, attempts are being
made to augment surface tumor cell antigens by prestimulation with growth factors, such as cytokines (11).
TREATMENT OF HEMATOLOGICAL MALIGNANCIES
Blood-Cell Cancers
Surface antigens on B- and T-cell lymphocytes are also
useful targets for the treatment of blood cell (hematopoietic)
malignancies, such as leukemias and lymphomas. These
antigens are also expressed at high levels on the surface of
various populations of malignant cells, but not on normal
tissues. With few barriers present to impede Mab binding,
hematologic malignancies are well suited to Mab-based
therapy. In recent years, several promising Mab-based
therapies for the treatment of hematologic malignancies
have been developed and either have already received U.S.
Food and Drug Administration (FDA) approval or are in
the advanced phases of clinical testing (12). The chimeric
antibody, rituxan (rituximAb, Genentech, San Francisco,
CA) was among the first Mabs awarded Food and Drug
Administration approval for the treatment of non-Hodgkins lymphoma (13,14). This chimeric (humanmouse)
antibody binds CD20, a cell surface antigen expressed on
mature B lymphocytes and over 90% of non-Hodgkins
lymphoma cells, but not on hematopoetic progenitor or
stem cells. Rituxan has proven to be well tolerated and
effective in the treatment of non-Hodgkins lymphoma
either by itself, or in combination with traditional chemotherapy, particularly in patients who are refractory to
other types of therapy (15). Campath-1 (alemtuzumAb, Ilex
Oncology, San Antonio, TX) is another antibody that has
also received FDA approval for the treatment of patients
suffering from chronic lymphocytic leukemia. A third Mab
to receive FDA approval for the treatment of hematologic
malignancies is the chimeric Mab, mylotarg (gemtuzumAb
ozogamicin, Wyeth-Ayerst Laboratories, Philadelphia,
PA). This antibody targets the CD33 antigen expressed
on myeloid (white cells) precursors and leukemic cells, and
is absent from normal tissues and pluripotent hematopoetic (blood-cell producing) stem cells.
TREATMENT OF SOLID TUMORS

In comparison to the management of hematologic malignancies, successful treatment of solid tumors with Mabs has
proven more elusive; however, some significant therapeutic
benefits have been achieved. Herceptin (trastuzumAb, Genentech) is a humanized antibody that has received FDA
approval for the treatment of metastatic breast cancer. This
Mab recognizes an extracellular domain of the HER-2 protein. Clinical trials with herceptin have shown it to be well
tolerated both as a single agent for second or third line
therapy, or in combination with chemotherapeutic agents as
a first line of therapy. Combination therapy resulted in a

25% improvement of overall survival in patients with
tumors that overexpress HER-2, and that are refractory
to other forms of treatment (16).
The antiepithelial cellular adhesion tumors Mab molecule, Panorex (eclrecolomAb, GlaxoSmith-Kline, United
Kingdom), is another Mab -based therapy that is currently
being used for the treatment of colorectal cancer. Panorex
has shown tangible benefit for cancer patients and has
received approval in Germany for the treatment of
advanced colorectal cancer. Like other Mabs used for the
treatment of solid tumors, Panorex has proven more efficacious in the treatment of micrometastatic lesions and
minimal residual disease in comparison to bulky tumor
masses (17).
The failure of Mabs in the treatment of bulky lesions is
primarily attributable to the low level of injected Mabs that
actually reaches its target within a sizable solid-tumor
mass. Studies using radiolabeled Mabs suggested that only
a very small percentage of the original injected antibody
dose, 0.010.1/g of tumor tissue, will ever reach target
antigens within a solid tumor (18). This low level of binding
is due to the series of barriers confronted by an administered Mab en route to antigens expressed on the surface of
tumor cells.
ELICITING ANTITUMOR RESPONSES

After successfully negotiating the gauntlet of obstacles
obstructing access to the target cells within a tumor, a
therapeutic Mab must still be capable of eliciting a potent
antitumor response. Although it is often ambiguous as to
the exact mechanisms by which a particular Mab may
mediate an antitumor response, both direct and indirect
mechanisms can potentially be involved.
Antibodies of the IgG1 and IgG3 isotypes can support
effector functions of both antibody-dependent cellmediated cytotoxicity and complement-dependent cytotoxicity. Antibody-dependent cell-mediated cytotoxicity is triggered by interaction between the Fc region of a cell-bound
antibody and Fc receptors on immune effector cells such as
neutrophils, macrophages, and natural killer cells. This
mechanism is critical for the antitumor effects of several
therapeutic Mabs.
Many early studies showed that murine Mabs had
limited potential to elicit a potent antitumor response,
because the murine Fc regions are less efficient at recruiting human effector cells than their human counterparts.
This problem has been largely alleviated by the use of
chimeric and humanized antibodies. Genetic engineering
techniques have also been used to improve the immunologic effects of therapeutic Mabs by altering antibody shape
and size, increasing the valency (bonds of affinity) of Mabs,
and creating bifunctional antibodies with two antigenic
receptors, one to a tumor antigen and another to an effector
cell to increase efficiency of antibody-dependent cellmediated cytotoxicity (19).
In addition to immunologic effects, Mabs can induce
antitumor effects by a variety of direct mechanisms, including the induction of apoptosis (programmed cell
death) (20), or the prevention of soluble growth factors from

binding their cognate receptors, such as epidermal growth
factor (EGF-R) (21) and HER-2 (22). Additionally, Mabs
can also be engineered to deliver a cytotoxic agent directly
to the tumor. This offers the potential to combine the
biological effects of Mabs with the additional effect of a
targeted cytotoxic response. The anti-CD33 Mabs, mylotarg, is one such antibody. Combined with the cytotoxic
agent, calichaemicin, mylotarg has been reported to be
relatively well tolerated, and effective in the treatment
of chronic lymphocytic leukemia (23). Antibodies can also
be engineered to deliver ionizing radiation directly to
tumor cells. Mabs have been conjugated to both a- and
b-particle emitting radionuclides (24). Clinical trials in
humans also portend the promise of radiolabeled Mabs
for the treatment of cancer. In a recent phase III randomized study, patients with relapsed or refractory nonHodgkins lymphoma were treated with yttrium-90 and
iodine-131 labeled Mabs targeting the CD20 antigen (ibrituximomAb, tiuxetan, and tositumomAb, respectively).
Patients treated with these radiolabeled Mabs showed a
statistically significant increase in overall response compared with those treated with an unlabeled version of the
Mab (rituximAb) (25).
OTHER USES FOR MONOCLONAL ANTIBODIES
Proteomics
After having sequenced the entire human genome, the
current task is to understand the proteome by identification and quantification of all proteins in a given sample. So
far, DNA microarrays have been employed to detect the
transcription level [production of messenger ribonucleic
acid (mRNA)] of genes in cells. However, it has been found
that there is no stringent correlation between transcription
level and protein abundance. Furthermore, the status of a
protein in terms of its modification and structure cannot be
determined by DNA microarrays. To solve this problem,
antibody microarrays are envisioned to replace DNA
microarrays in proteome research. These arrays consist
of a multitude of different antibodies that are immobilized
on a solid support and allow characterization of the protein
repertoire of a given sample. However, the production of
such antibody microarrays and its application require the
provision of highly specific and stable antibodies, possessing high affinity and showing no cross-reactivity. Protein
and antibody microarrays can be made to encompass as
many as 10,000 samples on a chip within the dimensions of
a microscope slide (26).
Monoclonal Antibodies in the Food Industry
Monoclonal antibodies are being used in the wine industry.
Odors sometimes observed in spoiled food or corked wines
are often the result of microbes present in the wood packaging materials. However, this phenomenon has also been
observed in bottled water, suggesting that there may be
secondary contaminants, such as residues of pesticides
that can affect the quality of any packaged food or beverage. To further the quality assurance of products in the
607
wine industry, a project is being carried out to raise antibodies against a TCA molecule (2,4,6-trichcholoanisole)
that is thought to be present in cork stoppers and is
responsible for the musty taste in wine. The ELISA assay
will be employed to detect trace amounts of the contaminating molecule. Also an immunosensor will be used to
electrochemically detect the antibody levels present (27).
Monoclonal antibodies have also been developed against
the vegetative cells and spores of Bacillus cereus (28). This
bacterium seems to be implicated in food poisoning and is
also responsible for food spoilage. It is impossible for the
food industry to exclude B. cereus from its products because
B. cereus cells can survive heat processing and can grow in
foods kept at refrigerated storage conditions. Two different
antibodies were developed. One was used as a specific
capture antibody to destroy the bacterium; the other as
a detector antibody that would simply identify the presence
of B. cereus. The ELISA assay was used to detect and
quantify the vegetative cells of this pathogenic organism.
Potato cyst nematodes are pests that destroy the potato
food crops. Monoclonal antibodies are being used to assist
in the development of the plants resistance to the nematode (29). Recombinant plant monoclonal antibodies have
been engineered to protect poultry against coccidosis infections (30).
Monoclonal Antibodies and Bioterrorism
The same plant biotechnology described above is being
developed to create strategic reserves of vaccines and
antibodies for infectious agents that could be used in
biowarfare. Multiple genes can be engineered in plants
intended to provide prolonged immunity against new
strains of pathogens that have different mechanisms of
action. With this technology, every plant cell will produce
the signature protein of a particular biowarfare agent.
That protein, in turn, will trigger an immune response
in a person who consumes the plant material in an unprocessed or lightly processed form, but it will not cause the
disease. These antibodies can prevent infection on surface
areas, including nasal passages; clear infectious organisms
from the body; identify foreign organisms for destruction;
and neutralize and remove toxins. Among the diseasecausing substances are several potential bioterrorism
agents, such as the botulism toxin, anthrax, Ebola virus,
plague, and ricin, a poisonous protein found in the seeds of
the castor oil plant. Vaccines for anthrax (Bacillus anthracis) and bubonic and pneumonic plague (Yersinia pestis),
two potentially deadly diseases that can be delivered as airborne agents, are being developed. Preliminary data
predicts success in using these plant-derived vaccines (31).
CONCLUDING REMARKS
Antibodies, monoclonal antibodies and antibody derivatives constitute 20 % of biopharmaceutical products currently in development. Antibodies represent an important
and growing class of biotherapeutics. Progress in antibody
engineering has allowed the manipulation of the basic
antibody structure into its minimal essential functions,
and multiple methodologies have emerged for raising
608
and tailoring specificity and functionality. The myriad of

monoclonal antibody structures that can be designed and
obtained in different formats from various production systems (bacterial, mammalian, and plants) represents a
challenge for the recovery and purification of novel immunotherapeutics (3). However, the general use in clinical
practice of antibody therapeutics is dependent not only on
the availability of products with required efficacy but also
on the costs of therapy. As a rule, a significant percentage
(5080%) of the total manufacturing cost of a therapeutic
antibody is incurred during downstream processing. The
critical challenges posed by the production of novel antibody therapeutics include improving process economics
and efficiency to reduce costs, and fulfilling increasingly
demanding quality criteria for FDA approval.
BIBLIOGRAPHY
1. Janeway CA JR, Travers P, Walport M, Shlomchik M. Immunobiology. The Immune System in Health and Disease. 6th
ed. Churchill Livingstone.
2. Kohler G, Milstein C. Continuous cultures of fused cells
secreting antibody of predefined specificity. Nature (London)
1975;256:495497.
3. Roque AC, Lowe CR, Taipa MA. Antibodies and genetically
engineered related molecules: Production and purification.
Biotechnol Prog 2004;20:639654.
4. Kretzschmar T, von Ruden T. Antibody discovery: Phage
display. Curr Opin Biotechnol 2002;13:598602.
5. Fischer R, Twyman RM, Schillberg S. Production of antibodies in
plants and their use for global health. Vaccine 2003; 21:820825.
6. Chen HR. et al. Humanized anti-CD25 monoclonal antibody
for prophylaxis of graft-vs-host disease (GVHD) in haploidentical bone marrow transplantation without ex vivo T-cell
depletion. Exp. Hematol 2003;31:10191025.
7. Hartmann TB, Bazhin AV, Schadendorf D, Eichmuller SB.
SEREX identification of new tumor antigens linked to melanoma-associated retinopathy. Int. J. Cancer 2005;10;114:88
93. http://www.licr.org/SEREX.html.
8. Davis DW. et al. Regional effects of an antivascular endothelial growth factor receptor monoclonal antibody on receptor
phosphorylation and apoptosis in human 253J B-V bladder
cancer xenografts. Cancer Res 2004;64:46014610.
9. Christiansen J, Rajasekaran AK. Biological impediments to
monoclonal antibodybased cancer immunotherapy. Mol
Cancer Ther 2004;3:14931501.
10. Koral KF. Update on hybrid conjugate-view SPECT tumor
dosimetry and response in 131I-tositumomab therapy of
previously untreated lymphoma patients. J Nucl Med
2003;44: 457464.
11. Villa AM, Berman B. Immunomodulators for skin cancer. J
Drugs Dermatol 2004;3:533539.
12. Von Mehren M, Adams GP, Weiner LM. Monoclonal antibody
therapy for cancer. Annu Rev Med 2003;54:343369.
13. Leget GA, Czuczman MS. Use of rituximAb, the new FDAapproved antibody. Curr Opin Oncol 1998;10:548551.
14. McLaughlin P. et al. RituximAb chimeric anti-CD20 monoclonal antibody therapy for relapsed indolent lymphoma:
Half of patients respond to a four-dose treatment program.
J Clin Oncol 1998;16:28252833.
15. Leyland-Jones B. TrastuzumAb: hopes and realities. Lancet
Oncol 2002;3:137144.
16. Riethmuller G. et al. Monoclonal antibody therapy for resected
Dukes C colorectal cancer: Seven-year outcome of a multicenter
randomized trial. J Clin Oncol 1998;16:17881794.
17. Mellstedt H, et al. The therapeutic use of monoclonal antibodies

in colorectal carcinoma. Semin Oncol 1991;18:462477.
18. Khawli LA, Miller GK, Epstein AL. Effect of seven new
vasoactive immunoconjugates on the enhancement of monoclonal antibody uptake in tumors. Cancer 1994;73:824831.
19. Zeidler R, et al. The Fc-region of a new class of intact
bispecific antibody mediates activation of accessory cells
and NK cells and induces direct phagocytosis of tumour cells.
Br J Cancer 2000;83:261266.
20. Trauth BC, et al. Monoclonal antibody-mediated tumor regression by induction of apoptosis. Science 1989;245:301305.
21. Yang XD, et al. Eradication of established tumors by a fully
human monoclonal antibody to the epidermal growth factor
receptor without concomitant chemotherapy. Cancer Res
1999;59:12361243.
22. Agus DB, et al. Targeting ligand-activated ErbB2 signaling
inhibits breast and prostate tumor growth. Cancer Cell
2002;2:127137.
23. Estey EH, et al. Experience with gemtuzumAb ozogamycin
(mylotarg) and all-trans retinoic acid in untreated acute
promyelocytic leukemia. Blood 2002;99:42224224.
24. Bander NH, et al. Targeted systemic therapy of prostate
cancer with a monoclonal antibody to prostate-specific membrane antigen. Semin Oncol 2003;30:667676.
25. Witzig TE, et al. Randomized controlled trial of yttrium-90
labeled ibritumomAb tiuxetan radioimmunotherapy versus
rituximAb immunotherapy for patients with relapsed or
refractory low-grade, follicular, or transformed B-cell nonHodgkins lymphoma. J Clin Oncol 2002;20:24532463.
26. Angenendt P, et al. Seeing better through a MIST: evaluation
of monoclonal recombinant antibody fragments on microarrays. Anal Chem 2004;76:29162921.
27. Sanvicens N, Varela B, Marco MP. Determination of 2,4,6trichloroanisole as the responsible agent for the musty odor
in foods. 2. Immunoassay evaluation. J Agric Food Chem
2003;51: 39323939.
28. Charni N, et al. Production and characterization of monoclonal antibodies against vegetative cells of Bacillus cereus.
Appl Environ Microbiol 2000;66:22782281.
29. Fioretti L, et al. Monoclonal antibodies reactive with secretedexcreted products from the amphids and the cuticle surface of
Globodera pallida affect nematode movement and delay invasion of potato roots. Int J Parasitol 2002;32:17091718.
30. Refega S, et al. Production of a functional chicken singlechain variable fragment antibody derived from caecal tonsils
B lymphocytes against macrogamonts of Eimeria tenella. Vet
Immunol Immunopathol 2004;97:219230.
31. Petrenko VA, Sorokulova IB. Detection of biological threats.
A challenge for directed molecular evolution. J Microbiol
Methods 2004;58:147168.
See also BORON
NEUTRON CAPTURE THERAPY; IMMUNOTHERAPY.
MOSFET. See ION-SENSITIVE FIELD-EFFECT TRANSISTORS.

MRI.
See MAGNETIC
RESONANCE IMAGING.
MUSCLE ELECTRICAL ACTIVITY. See

ELECTROMYOGRAPHY.
MUSCLE TESTING, REHABILITATION AND. See

REHABILITATION
AND MUSCLE TESTING.
MUSCULOSKELETAL DISABILITIES.
REHABILITATION,
ORTHOTICS FOR.
See

Enciclopedia de Dispositivos Médicos e Instrumentación

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Enciclopedia de Dispositivos Médicos e Instrumentación

Hochgeladen von

Copyright:

Verfügbare Formate

ENCYCLOPEDIA OF

MEDICAL DEVICES AND

ENCYCLOPEDIA OF MEDICAL DEVICES AND INSTRUMENTATION, SECOND EDITION

MEDICAL DEVICES AND

The Encyclopedia of Medical Devices and Instrumentation is available online at

A John Wiley & Sons, Inc., Publication

Library of Congress Cataloging-in-Publication Data:

Printed in the United States of America

BERSANO-BEGEY, TOMMASO, University of Michigan, Ann

CAPPELLERI, JOSEPH C., Pfizer Inc., Groton, Connecticut,

DAMBRA, MICHAEL N., Harvard Medical School, Cambridge,

DYSON, MARY, Herts, United Kingdom, Heat and Cold,

GHARIEB, R.R., Infinite Biomedical Technologies, Baltimore,

HOLDER, GRAHAM E., Moorfields Eye Hospital, London,

KATZ, J. LAWRENCE, University of Missouri-Kansas City,

MADSEN, MARK T., University of Iowa, Iowa City, Iowa,

MURPHY, WILLIAM L., University of Wisconsin, Madison,

PURDY, JAMES A., UC Davis Medical Center, Sacramento,

SCOPE, KENNETH, Northwestern University, Chicago,

SPELMAN, FRANCIS A., University of Washington, Cochlear

WANG, LE YI, Wayne State University, Detroit, Michigan,

YENI, YENER N., Henry Ford Hospital, Detroit, Michigan,

ZAIDER, MARCO, Memorial Sloan Kettering Cancer Center,

ALLOYS, SHAPE MEMORY

LINEAR VARIABLE DIFFERENTIAL TRANSFORMERS

PICTURE ARCHIVING AND COMMUNICATION

ABBREVIATIONS AND ACRONYMS

Association for the Advancement of

Advanced life support; Amyotropic

ABBREVIATIONS AND ACRONYMS

Closed circuit television system

ABBREVIATIONS AND ACRONYMS

Early burn scar

ABBREVIATIONS AND ACRONYMS

Electron spectroscopy for chemical

Gas chromatography; Guanine-cytosine

ABBREVIATIONS AND ACRONYMS

Heart rate; High-resolution

International Medical Informatics

ABBREVIATIONS AND ACRONYMS

Left coronary cusp

Mean diastolic aortic pressure

ABBREVIATIONS AND ACRONYMS

Neutron activation analysis

ABBREVIATIONS AND ACRONYMS

Prolonged ultraviolet-A radiation

ABBREVIATIONS AND ACRONYMS

Surgical isolation barrier system

ABBREVIATIONS AND ACRONYMS

Ultra high molecular weight polyethylene

CONVERSION FACTORS AND UNIT SYMBOLS

Derived Units and Other Acceptable Units

CONVERSION FACTORS AND UNIT SYMBOLS

heat capacity, entropy

power density, heat flux density,

watt per square meter

CONVERSION FACTORS AND UNIT SYMBOLS

In addition, there are 16 prefixes used to indicate order of magnitude, as follows:

Although hecto, deka, deci, and centi are

CONVERSION FACTORS TO SI UNITS

CONVERSION FACTORS AND UNIT SYMBOLS

CONVERSION FACTORS AND UNIT SYMBOLS

HYDROCEPHALUS, TOOLS FOR DIAGNOSIS

CSF Circulation. Unlike other organ systems, the brain